fix(moonshot): backfill reasoning_content on assistant tool-call replay messages (#92396)

Moonshot/Kimi requires reasoning_content on all assistant tool-call messages
when thinking is enabled. After LCM compaction, cross-model fallback, or
session repair, the replayed history may be missing this field, causing a
400 error from the Moonshot API.

Backfill an empty string to satisfy the API schema contract without
fabricating semantic reasoning content. Follows the same provider-owned
backfill pattern already used by Kimi Coding (extensions/kimi-coding/stream.ts)
and DeepSeek V4 (provider-stream-shared.ts).

Fixes #71491

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

.github/workflows/mantis-discord-status-reactions.yml

@@ -437,8 +437,17 @@ jobs:
             echo "::warning::Could not generate motion-trimmed desktop previews; continuing with screenshots and full MP4 links."
           fi
 
-          baseline_status="$(jq -r '.scenarios[0].status' "$root/baseline/discord-qa-summary.json")"
-          candidate_status="$(jq -r '.scenarios[0].status' "$root/candidate/discord-qa-summary.json")"
+          read_discord_status_reaction_status() {
+            local lane="$1"
+            if [[ -f "$root/$lane/qa-evidence.json" ]]; then
+              jq -r '.entries[0].result.status' "$root/$lane/qa-evidence.json"
+              return
+            fi
+            jq -r '.scenarios[0].status' "$root/$lane/discord-qa-summary.json"
+          }
+
+          baseline_status="$(read_discord_status_reaction_status baseline)"
+          candidate_status="$(read_discord_status_reaction_status candidate)"
 
           jq -n \
             --arg baseline_status "$baseline_status" \

.github/workflows/mantis-discord-thread-attachment.yml

@@ -451,8 +451,17 @@ jobs:
 
           capture_candidate_discord_web
 
-          baseline_status="$(jq -r '.scenarios[] | select(.id == "discord-thread-reply-filepath-attachment") | .status' "$root/baseline/discord-qa-summary.json")"
-          candidate_status="$(jq -r '.scenarios[] | select(.id == "discord-thread-reply-filepath-attachment") | .status' "$root/candidate/discord-qa-summary.json")"
+          read_discord_thread_attachment_status() {
+            local lane="$1"
+            if [[ -f "$root/$lane/qa-evidence.json" ]]; then
+              jq -r '.entries[] | select(.test.id == "discord-thread-reply-filepath-attachment") | .result.status' "$root/$lane/qa-evidence.json"
+              return
+            fi
+            jq -r '.scenarios[] | select(.id == "discord-thread-reply-filepath-attachment") | .status' "$root/$lane/discord-qa-summary.json"
+          }
+
+          baseline_status="$(read_discord_thread_attachment_status baseline)"
+          candidate_status="$(read_discord_thread_attachment_status candidate)"
           comparison_status="fail"
           if [[ "$baseline_status" == "fail" && "$candidate_status" == "pass" ]]; then
             comparison_status="pass"

.github/workflows/mantis-telegram-live.yml

@@ -445,8 +445,8 @@ jobs:
           telegram_exit=$?
           set -e
 
-          if [[ ! -f "$root/telegram-qa-summary.json" ]]; then
-            echo "Telegram live QA did not produce a summary." >&2
+          if [[ ! -f "$root/qa-evidence.json" && ! -f "$root/telegram-qa-summary.json" ]]; then
+            echo "Telegram live QA did not produce an evidence summary." >&2
             exit "$telegram_exit"
           fi
           echo "telegram_exit=${telegram_exit}" >> "$GITHUB_OUTPUT"

.github/workflows/openclaw-live-and-e2e-checks-reusable.yml

@@ -1748,6 +1748,7 @@ jobs:
             anthropic) require_any Anthropic ANTHROPIC_API_KEY ANTHROPIC_API_KEY_OLD ANTHROPIC_API_TOKEN ;;
             google) require_any Google GEMINI_API_KEY GOOGLE_API_KEY ;;
             minimax) require_any MiniMax MINIMAX_API_KEY ;;
+            moonshot) require_any Moonshot MOONSHOT_API_KEY KIMI_API_KEY ;;
             openai) require_any OpenAI OPENAI_API_KEY ;;
             opencode-go) require_any OpenCode OPENCODE_API_KEY OPENCODE_ZEN_API_KEY ;;
             openrouter) require_any OpenRouter OPENROUTER_API_KEY ;;
@@ -1836,7 +1837,7 @@ jobs:
         run: |
           set -euo pipefail
 
-          all_providers=(anthropic google minimax openai opencode-go openrouter xai zai fireworks)
+          all_providers=(anthropic google minimax moonshot openai opencode-go openrouter xai zai fireworks)
 
           normalize_provider() {
             local value="${1,,}"
@@ -1922,6 +1923,7 @@ jobs:
               anthropic) require_any Anthropic ANTHROPIC_API_KEY ANTHROPIC_API_KEY_OLD ANTHROPIC_API_TOKEN ;;
               google) require_any Google GEMINI_API_KEY GOOGLE_API_KEY ;;
               minimax) require_any MiniMax MINIMAX_API_KEY ;;
+              moonshot) require_any Moonshot MOONSHOT_API_KEY KIMI_API_KEY ;;
               openai) require_any OpenAI OPENAI_API_KEY ;;
               opencode-go) require_any OpenCode OPENCODE_API_KEY OPENCODE_ZEN_API_KEY ;;
               openrouter) require_any OpenRouter OPENROUTER_API_KEY ;;

.github/workflows/plugin-npm-release.yml

@@ -288,6 +288,7 @@ jobs:
         env:
           NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}
           NPM_TOKEN: ${{ secrets.NPM_TOKEN }}
+          OPENCLAW_NPM_PUBLISH_AUTH_MODE: trusted-publisher
         run: bash scripts/plugin-npm-publish.sh --publish "${{ matrix.plugin.packageDir }}"
 
       - name: Verify published runtime

Dockerfile

@@ -116,11 +116,19 @@ RUN pnpm_config_verify_deps_before_run=false pnpm canvas:a2ui:bundle || \
      echo "/* A2UI bundle unavailable in this build */" > extensions/canvas/src/host/a2ui/a2ui.bundle.js && \
      echo "stub" > extensions/canvas/src/host/a2ui/.bundle.hash && \
      rm -rf vendor/a2ui apps/shared/OpenClawKit/Tools/CanvasA2UI)
-RUN NODE_OPTIONS=--max-old-space-size=8192 pnpm_config_verify_deps_before_run=false pnpm build:docker
+RUN if printf '%s\n' "$OPENCLAW_EXTENSIONS" | tr ',' ' ' | tr ' ' '\n' | grep -qx 'qa-lab'; then \
+      export OPENCLAW_BUILD_PRIVATE_QA=1 OPENCLAW_ENABLE_PRIVATE_QA_CLI=1; \
+    fi && \
+    NODE_OPTIONS=--max-old-space-size=8192 pnpm_config_verify_deps_before_run=false pnpm build:docker
 # Force pnpm for UI build (Bun may fail on ARM/Synology architectures)
 ENV OPENCLAW_PREFER_PNPM=1
 RUN pnpm_config_verify_deps_before_run=false pnpm ui:build
-RUN pnpm_config_verify_deps_before_run=false pnpm qa:lab:build
+RUN if printf '%s\n' "$OPENCLAW_EXTENSIONS" | tr ',' ' ' | tr ' ' '\n' | grep -qx 'qa-lab'; then \
+      pnpm_config_verify_deps_before_run=false pnpm qa:lab:build && \
+      mkdir -p dist/extensions/qa-lab/web && \
+      rm -rf dist/extensions/qa-lab/web/dist && \
+      cp -R extensions/qa-lab/web/dist dist/extensions/qa-lab/web/dist; \
+    fi
 
 # Prune dev dependencies, omitted plugin runtime packages, and build-only
 # metadata before copying runtime assets into the final image.

docs/.generated/plugin-sdk-api-baseline.sha256

@@ -1,2 +1,2 @@
-8a2769df428906990ee0d1bf8b0423f2a099b053c64c816d092ff84d61e11633  plugin-sdk-api-baseline.json
-28b798973f3fb2a5b33ccbb6e3c1ac0453fa234a3a1c6cdc27935c27639bd104  plugin-sdk-api-baseline.jsonl
+2c783beea6b3cda3d79060739a923f9f39e7e8b5942123dd6b08a09143a587ca  plugin-sdk-api-baseline.json
+0b33af2cffb42abb46682fb71c8f214da220793f13d10a34d332e75ff99e8ce9  plugin-sdk-api-baseline.jsonl

docs/concepts/model-providers.md

@@ -368,6 +368,7 @@ Kimi K2 model IDs:
 [//]: # "moonshot-kimi-k2-model-refs:start"
 
 - `moonshot/kimi-k2.6`
+- `moonshot/kimi-k2.7-code`
 - `moonshot/kimi-k2.5`
 - `moonshot/kimi-k2-thinking`
 - `moonshot/kimi-k2-thinking-turbo`

docs/concepts/qa-e2e-automation.md

@@ -374,7 +374,7 @@ The implicit default set always covers canary, mention gating, native command re
 Output artifacts:
 
 - `telegram-qa-report.md`
-- `telegram-qa-summary.json` - includes per-reply RTT (driver send → observed SUT reply) starting with the canary.
+- `qa-evidence.json` - evidence entries for the live transport checks, including profile, coverage, provider, channel, artifacts, result, and RTT fields.
 - `telegram-qa-observed-messages.json` - bodies redacted unless `OPENCLAW_QA_TELEGRAM_CAPTURE_CONTENT=1`.
 
 Package RTT comparison uses the same Telegram credential contract while keeping
@@ -447,7 +447,7 @@ pnpm openclaw qa discord \
 Output artifacts:
 
 - `discord-qa-report.md`
-- `discord-qa-summary.json`
+- `qa-evidence.json` - evidence entries for the live transport checks.
 - `discord-qa-observed-messages.json` - bodies redacted unless `OPENCLAW_QA_DISCORD_CAPTURE_CONTENT=1`.
 - `discord-qa-reaction-timelines.json` and `discord-status-reactions-tool-only-timeline.png` when the status-reaction scenario runs.
 
@@ -495,7 +495,7 @@ Scenarios (`extensions/qa-lab/src/live-transports/slack/slack-live.runtime.ts`):
 Output artifacts:
 
 - `slack-qa-report.md`
-- `slack-qa-summary.json`
+- `qa-evidence.json` - evidence entries for the live transport checks.
 - `slack-qa-observed-messages.json` - bodies redacted unless `OPENCLAW_QA_SLACK_CAPTURE_CONTENT=1`.
 - `approval-checkpoints/` - only when Mantis sets
   `OPENCLAW_QA_SLACK_APPROVAL_CHECKPOINT_DIR`; contains checkpoint JSON,
@@ -740,7 +740,7 @@ poll and upload-file coverage run through deterministic gateway `poll` and
 Output artifacts:
 
 - `whatsapp-qa-report.md`
-- `whatsapp-qa-summary.json`
+- `qa-evidence.json` - evidence entries for the live transport checks.
 - `whatsapp-qa-observed-messages.json` - bodies redacted unless `OPENCLAW_QA_WHATSAPP_CAPTURE_CONTENT=1`.
 
 ### Convex credential pool

docs/docs.json

@@ -1374,7 +1374,6 @@
                 "pages": [
                   "clawhub/cli",
                   "clawhub/publishing",
-                  "clawhub/plugin-validation-fixes",
                   "clawhub/skill-format",
                   "clawhub/auth",
                   "clawhub/telemetry",

docs/plugins/codex-native-plugins.md

@@ -200,12 +200,11 @@ enabled.
 
 OpenClaw sets app-level `destructive_enabled` from the effective global or
 per-plugin `allow_destructive_actions` policy and lets Codex enforce
-destructive tool metadata from its native app tool annotations. `true` and
-`"on-request"` both set `destructive_enabled: true`; `false` sets it false. The
-`_default` app config is disabled with `open_world_enabled: false`. Enabled
-plugin apps are emitted with `open_world_enabled: true`; OpenClaw does not
-expose a separate plugin open-world policy knob and does not maintain
-per-plugin destructive tool-name deny lists.
+destructive tool metadata from its native app tool annotations. The `_default`
+app config is disabled with `open_world_enabled: false`. Enabled plugin apps
+are emitted with `open_world_enabled: true`; OpenClaw does not expose a separate
+plugin open-world policy knob and does not maintain per-plugin destructive
+tool-name deny lists.
 
 Tool approval mode is automatic by default for plugin apps so non-destructive
 read tools can run without a same-thread approval UI. Destructive tools remain
@@ -222,9 +221,6 @@ plugins, while unsafe schemas and ambiguous ownership still fail closed:
 - When policy is `false`, OpenClaw returns a deterministic decline.
 - When policy is `true`, OpenClaw auto-accepts only safe schemas it can map to
   an approval response, such as a boolean approve field.
-- When policy is `"on-request"`, OpenClaw exposes destructive plugin actions to
-  Codex but turns ownership-proven MCP approval elicitations into OpenClaw
-  plugin approvals before returning the Codex approval response.
 - Missing plugin identity, ambiguous ownership, a missing turn id, a wrong turn
   id, or an unsafe elicitation schema declines instead of prompting.
 
@@ -272,8 +268,8 @@ Codex thread bindings keep the app config they started with until OpenClaw
 establishes a new harness session or replaces a stale binding.
 
 **Destructive action is declined:** check the global and per-plugin
-`allow_destructive_actions` values. Even when policy is true or `"on-request"`,
-unsafe elicitation schemas and ambiguous plugin identity still fail closed.
+`allow_destructive_actions` values. Even when policy is true, unsafe elicitation
+schemas and ambiguous plugin identity still fail closed.
 
 ## Related
 

docs/providers/moonshot.md

@@ -22,6 +22,7 @@ Moonshot and Kimi Coding are **separate providers**. Keys are not interchangeabl
 | Model ref                         | Name                   | Reasoning | Input       | Context | Max output |
 | --------------------------------- | ---------------------- | --------- | ----------- | ------- | ---------- |
 | `moonshot/kimi-k2.6`              | Kimi K2.6              | No        | text, image | 262,144 | 262,144    |
+| `moonshot/kimi-k2.7-code`         | Kimi K2.7 Code         | Always on | text, image | 262,144 | 262,144    |
 | `moonshot/kimi-k2.5`              | Kimi K2.5              | No        | text, image | 262,144 | 262,144    |
 | `moonshot/kimi-k2-thinking`       | Kimi K2 Thinking       | Yes       | text        | 262,144 | 262,144    |
 | `moonshot/kimi-k2-thinking-turbo` | Kimi K2 Thinking Turbo | Yes       | text        | 262,144 | 262,144    |
@@ -30,11 +31,18 @@ Moonshot and Kimi Coding are **separate providers**. Keys are not interchangeabl
 [//]: # "moonshot-kimi-k2-ids:end"
 
 Bundled cost estimates for current Moonshot-hosted K2 models use Moonshot's
-published pay-as-you-go rates: Kimi K2.6 is $0.16/MTok cache hit,
+published pay-as-you-go rates: Kimi K2.7 Code is $0.19/MTok cache hit,
+$0.95/MTok input, and $4.00/MTok output; Kimi K2.6 is $0.16/MTok cache hit,
 $0.95/MTok input, and $4.00/MTok output; Kimi K2.5 is $0.10/MTok cache hit,
 $0.60/MTok input, and $3.00/MTok output. Other legacy catalog entries keep
 zero-cost placeholders unless you override them in config.
 
+Kimi K2.7 Code always uses native thinking. OpenClaw exposes only the `on`
+thinking state for this model and omits outbound `thinking` and
+`reasoning_effort` controls, as required by Moonshot. OpenClaw also omits
+sampling overrides that K2.7 fixes to provider defaults. Kimi K2.6 remains the
+onboarding default.
+
 ## Getting started
 
 Choose your provider and follow the setup steps.
@@ -109,6 +117,7 @@ Choose your provider and follow the setup steps.
           models: {
             // moonshot-kimi-k2-aliases:start
             "moonshot/kimi-k2.6": { alias: "Kimi K2.6" },
+            "moonshot/kimi-k2.7-code": { alias: "Kimi K2.7 Code" },
             "moonshot/kimi-k2.5": { alias: "Kimi K2.5" },
             "moonshot/kimi-k2-thinking": { alias: "Kimi K2 Thinking" },
             "moonshot/kimi-k2-thinking-turbo": { alias: "Kimi K2 Thinking Turbo" },
@@ -135,6 +144,15 @@ Choose your provider and follow the setup steps.
                 contextWindow: 262144,
                 maxTokens: 262144,
               },
+              {
+                id: "kimi-k2.7-code",
+                name: "Kimi K2.7 Code",
+                reasoning: true,
+                input: ["text", "image"],
+                cost: { input: 0.95, output: 4, cacheRead: 0.19, cacheWrite: 0 },
+                contextWindow: 262144,
+                maxTokens: 262144,
+              },
               {
                 id: "kimi-k2.5",
                 name: "Kimi K2.5",
@@ -288,7 +306,13 @@ Config lives under `plugins.entries.moonshot.config.webSearch`:
 
 <AccordionGroup>
   <Accordion title="Native thinking mode">
-    Moonshot Kimi supports binary native thinking:
+    Kimi K2.7 Code always uses native thinking. Moonshot requires clients to
+    omit the `thinking` field for this model, so OpenClaw exposes only `on` and
+    ignores stale `off` settings. K2.7 also fixes `temperature`, `top_p`, `n`,
+    `presence_penalty`, and `frequency_penalty`; OpenClaw omits configured
+    overrides for those fields.
+
+    Other Moonshot Kimi models support binary native thinking:
 
     - `thinking: { type: "enabled" }`
     - `thinking: { type: "disabled" }`
@@ -311,7 +335,7 @@ Config lives under `plugins.entries.moonshot.config.webSearch`:
     }
     ```
 
-    OpenClaw also maps runtime `/think` levels for Moonshot:
+    OpenClaw maps runtime `/think` levels for those models:
 
     | `/think` level       | Moonshot behavior          |
     | -------------------- | -------------------------- |
@@ -319,14 +343,16 @@ Config lives under `plugins.entries.moonshot.config.webSearch`:
     | Any non-off level    | `thinking.type=enabled`    |
 
     <Warning>
-    When Moonshot thinking is enabled, `tool_choice` must be `auto` or `none`. OpenClaw normalizes incompatible `tool_choice` values to `auto` for compatibility.
+    When Moonshot thinking is enabled, `tool_choice` must be `auto` or `none`. OpenClaw normalizes incompatible values to `auto`. This includes Kimi K2.7 Code, whose thinking mode cannot be disabled to preserve a pinned tool choice.
     </Warning>
 
     Kimi K2.6 also accepts an optional `thinking.keep` field that controls
     multi-turn retention of `reasoning_content`. Set it to `"all"` to keep full
     reasoning across turns; omit it (or leave it `null`) to use the server
     default strategy. OpenClaw only forwards `thinking.keep` for
-    `moonshot/kimi-k2.6` and strips it from other models.
+    `moonshot/kimi-k2.6` and strips it from other models. Kimi K2.7 Code
+    preserves full reasoning history by default while OpenClaw omits the entire
+    `thinking` field.
 
     ```json5
     {

docs/tools/thinking.md

@@ -35,7 +35,7 @@ title: "Thinking levels"
   - Google Gemini maps `/think adaptive` to Gemini's provider-owned dynamic thinking. Gemini 3 requests omit a fixed `thinkingLevel`, while Gemini 2.5 requests send `thinkingBudget: -1`; fixed levels still map to the closest Gemini `thinkingLevel` or budget for that model family.
   - MiniMax M2.x (`minimax/MiniMax-M2*`) on the Anthropic-compatible streaming path defaults to `thinking: { type: "disabled" }` unless you explicitly set thinking in model params or request params. This avoids leaked `reasoning_content` deltas from M2.x's non-native Anthropic stream format. MiniMax-M3 (and M3.x) is exempt: M3 emits proper Anthropic thinking blocks and returns empty content when thinking is disabled, so OpenClaw keeps M3 on the provider's omitted/adaptive thinking path.
   - Z.AI (`zai/*`) only supports binary thinking (`on`/`off`). Any non-`off` level is treated as `on` (mapped to `low`).
-  - Moonshot (`moonshot/*`) maps `/think off` to `thinking: { type: "disabled" }` and any non-`off` level to `thinking: { type: "enabled" }`. When thinking is enabled, Moonshot only accepts `tool_choice` `auto|none`; OpenClaw normalizes incompatible values to `auto`.
+  - Moonshot Kimi K2.7 Code (`moonshot/kimi-k2.7-code`) always thinks. Its profile exposes only `on`, and OpenClaw omits the outbound `thinking` field as required by Moonshot. Other `moonshot/*` models map `/think off` to `thinking: { type: "disabled" }` and any non-`off` level to `thinking: { type: "enabled" }`. When thinking is enabled, Moonshot only accepts `tool_choice` `auto|none`; OpenClaw normalizes incompatible values to `auto`.
 
 ## Resolution order
 

extensions/acpx/npm-shrinkwrap.json

@@ -224,9 +224,9 @@
       }
     },
     "node_modules/@esbuild/aix-ppc64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/aix-ppc64/-/aix-ppc64-0.28.0.tgz",
-      "integrity": "sha512-lhRUCeuOyJQURhTxl4WkpFTjIsbDayJHih5kZC1giwE+MhIzAb7mEsQMqMf18rHLsrb5qI1tafG20mLxEWcWlA==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/aix-ppc64/-/aix-ppc64-0.28.1.tgz",
+      "integrity": "sha512-Svl7tq8k/08+p6CXPpRjQ1fKX+1odH/BQbb48fV6fj3CWHhsoIOoY87w1oHXm0qEpkIK3ZfVgp0hed3XBXzXMQ==",
       "cpu": [
         "ppc64"
       ],
@@ -240,9 +240,9 @@
       }
     },
     "node_modules/@esbuild/android-arm": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/android-arm/-/android-arm-0.28.0.tgz",
-      "integrity": "sha512-wqh0ByljabXLKHeWXYLqoJ5jKC4XBaw6Hk08OfMrCRd2nP2ZQ5eleDZC41XHyCNgktBGYMbqnrJKq/K/lzPMSQ==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/android-arm/-/android-arm-0.28.1.tgz",
+      "integrity": "sha512-0k2F129Xdio1TdJfzJ8sy1Q47vUD2NnwdhiAf7drUN1EBTfPf4hsFCtmMgu/6m8JSzsBrlmVjudMBQqOfG8usQ==",
       "cpu": [
         "arm"
       ],
@@ -256,9 +256,9 @@
       }
     },
     "node_modules/@esbuild/android-arm64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/android-arm64/-/android-arm64-0.28.0.tgz",
-      "integrity": "sha512-+WzIXQOSaGs33tLEgYPYe/yQHf0WTU0X42Jca3y8NWMbUVhp7rUnw+vAsRC/QiDrdD31IszMrZy+qwPOPjd+rw==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/android-arm64/-/android-arm64-0.28.1.tgz",
+      "integrity": "sha512-34EGEbCIAgosYz6goLcopX6Mo7NyGv9tfwEM2/7Ce2VcVRk568iSvniGWcUXIy7wEDR1wzolcxcriFVrWYcwBg==",
       "cpu": [
         "arm64"
       ],
@@ -272,9 +272,9 @@
       }
     },
     "node_modules/@esbuild/android-x64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/android-x64/-/android-x64-0.28.0.tgz",
-      "integrity": "sha512-+VJggoaKhk2VNNqVL7f6S189UzShHC/mR9EE8rDdSkdpN0KflSwWY/gWjDrNxxisg8Fp1ZCD9jLMo4m0OUfeUA==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/android-x64/-/android-x64-0.28.1.tgz",
+      "integrity": "sha512-dbwY7ltSMDWsRatcRpCnES4F+im88OCUgGZjy52shC7GqHRE/cYlxNbB4Z4UpJswpcc4Qxd2oE/ufM0p61IKng==",
       "cpu": [
         "x64"
       ],
@@ -288,9 +288,9 @@
       }
     },
     "node_modules/@esbuild/darwin-arm64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/darwin-arm64/-/darwin-arm64-0.28.0.tgz",
-      "integrity": "sha512-0T+A9WZm+bZ84nZBtk1ckYsOvyA3x7e2Acj1KdVfV4/2tdG4fzUp91YHx+GArWLtwqp77pBXVCPn2We7Letr0Q==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/darwin-arm64/-/darwin-arm64-0.28.1.tgz",
+      "integrity": "sha512-TZbWkQY7kvTAXbXUT7uVACR5cMHsDiSz9z7ZKAX/RTq/WJEk3QyRr0wZpNhBDX+/0CtdqUIJlOiodQcta6tY3Q==",
       "cpu": [
         "arm64"
       ],
@@ -304,9 +304,9 @@
       }
     },
     "node_modules/@esbuild/darwin-x64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/darwin-x64/-/darwin-x64-0.28.0.tgz",
-      "integrity": "sha512-fyzLm/DLDl/84OCfp2f/XQ4flmORsjU7VKt8HLjvIXChJoFFOIL6pLJPH4Yhd1n1gGFF9mPwtlN5Wf82DZs+LQ==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/darwin-x64/-/darwin-x64-0.28.1.tgz",
+      "integrity": "sha512-zfdzgK9ACBNZLI/CyHTOx81SyNbM6YXn7rxSgX97VjyiPl9W1i4Ka4fgKECEoFCKGpvBj5qArWIGgQjOwkgskQ==",
       "cpu": [
         "x64"
       ],
@@ -320,9 +320,9 @@
       }
     },
     "node_modules/@esbuild/freebsd-arm64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/freebsd-arm64/-/freebsd-arm64-0.28.0.tgz",
-      "integrity": "sha512-l9GeW5UZBT9k9brBYI+0WDffcRxgHQD8ShN2Ur4xWq/NFzUKm3k5lsH4PdaRgb2w7mI9u61nr2gI2mLI27Nh3Q==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/freebsd-arm64/-/freebsd-arm64-0.28.1.tgz",
+      "integrity": "sha512-wG2EA8ENdEI0qhkSZMjfqrdY+ziCYCPMmtZjjIwOmXFjmyzEHn+UUxk5of+SYsjtfs3VpnlC7QLzSI5hY/rOAw==",
       "cpu": [
         "arm64"
       ],
@@ -336,9 +336,9 @@
       }
     },
     "node_modules/@esbuild/freebsd-x64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/freebsd-x64/-/freebsd-x64-0.28.0.tgz",
-      "integrity": "sha512-BXoQai/A0wPO6Es3yFJ7APCiKGc1tdAEOgeTNy3SsB491S3aHn4S4r3e976eUnPdU+NbdtmBuLncYir2tMU9Nw==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/freebsd-x64/-/freebsd-x64-0.28.1.tgz",
+      "integrity": "sha512-i7dZ9vQgnvSCzi/rYCXNgtF/U+eKZNJBzu3eTQbRgHnM7tNSizLOkRFAl3qzVc/Op/u5YkHHa4pf/3DOYHthLQ==",
       "cpu": [
         "x64"
       ],
@@ -352,9 +352,9 @@
       }
     },
     "node_modules/@esbuild/linux-arm": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/linux-arm/-/linux-arm-0.28.0.tgz",
-      "integrity": "sha512-CjaaREJagqJp7iTaNQjjidaNbCKYcd4IDkzbwwxtSvjI7NZm79qiHc8HqciMddQ6CKvJT6aBd8lO9kN/ZudLlw==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/linux-arm/-/linux-arm-0.28.1.tgz",
+      "integrity": "sha512-qVXBOHQS+d5Y722GwJzJUtOLlX7km3CraOaGormF1pDtPd2C/l1SHRPgjLunLGe51Sh5YYWKMFDyV4SxgMQYTQ==",
       "cpu": [
         "arm"
       ],
@@ -368,9 +368,9 @@
       }
     },
     "node_modules/@esbuild/linux-arm64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/linux-arm64/-/linux-arm64-0.28.0.tgz",
-      "integrity": "sha512-RVyzfb3FWsGA55n6WY0MEIEPURL1FcbhFE6BffZEMEekfCzCIMtB5yyDcFnVbTnwk+CLAgTujmV/Lgvih56W+A==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/linux-arm64/-/linux-arm64-0.28.1.tgz",
+      "integrity": "sha512-yHs+0uc8+nvEAfAfxrWQKK5peSNzBc4PegcMO0EJ2hT71uA7vB8Ihg2e77R2P7SG5uYjPbHlLLmve4LLLRCf0g==",
       "cpu": [
         "arm64"
       ],
@@ -384,9 +384,9 @@
       }
     },
     "node_modules/@esbuild/linux-ia32": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/linux-ia32/-/linux-ia32-0.28.0.tgz",
-      "integrity": "sha512-KBnSTt1kxl9x70q+ydterVdl+Cn0H18ngRMRCEQfrbqdUuntQQ0LoMZv47uB97NljZFzY6HcfqEZ2SAyIUTQBQ==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/linux-ia32/-/linux-ia32-0.28.1.tgz",
+      "integrity": "sha512-d1z4ZuP0ajrfz/FhGT4vv278rX8KnPPJx8i5+AtK7TYbx9Le9F1hyzurZpkEyjkGa9dUGhQow4C1NmeGvqxN2w==",
       "cpu": [
         "ia32"
       ],
@@ -400,9 +400,9 @@
       }
     },
     "node_modules/@esbuild/linux-loong64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/linux-loong64/-/linux-loong64-0.28.0.tgz",
-      "integrity": "sha512-zpSlUce1mnxzgBADvxKXX5sl8aYQHo2ezvMNI8I0lbblJtp8V4odlm3Yzlj7gPyt3T8ReksE6bK+pT3WD+aJRg==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/linux-loong64/-/linux-loong64-0.28.1.tgz",
+      "integrity": "sha512-M5sRjUVZrkm1OAPR3dlOYzNmN+loZKGVi1VUQGrwuqLcbR6qeAz+famMhjASeH3YVKvZz+zT1jlh/keC3Rj/lg==",
       "cpu": [
         "loong64"
       ],
@@ -416,9 +416,9 @@
       }
     },
     "node_modules/@esbuild/linux-mips64el": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/linux-mips64el/-/linux-mips64el-0.28.0.tgz",
-      "integrity": "sha512-2jIfP6mmjkdmeTlsX/9vmdmhBmKADrWqN7zcdtHIeNSCH1SqIoNI63cYsjQR8J+wGa4Y5izRcSHSm8K3QWmk3w==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/linux-mips64el/-/linux-mips64el-0.28.1.tgz",
+      "integrity": "sha512-mRObBZeHh2OxcBFPWE/FjylkRgZdYuiTR3vaTozquCGOH14iP9oN4x4Ge81CoIDYQrXmIxpFumJBu5MtZpnQJQ==",
       "cpu": [
         "mips64el"
       ],
@@ -432,9 +432,9 @@
       }
     },
     "node_modules/@esbuild/linux-ppc64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/linux-ppc64/-/linux-ppc64-0.28.0.tgz",
-      "integrity": "sha512-bc0FE9wWeC0WBm49IQMPSPILRocGTQt3j5KPCA8os6VprfuJ7KD+5PzESSrJ6GmPIPJK965ZJHTUlSA6GNYEhg==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/linux-ppc64/-/linux-ppc64-0.28.1.tgz",
+      "integrity": "sha512-slScBsMAb3GFDcdrCgLwZtPYRoH2H/youv10QiZyRjmsP48fznoveWytSgCI/R0ZcUgpc0ZhIUEx6LHts8yrfQ==",
       "cpu": [
         "ppc64"
       ],
@@ -448,9 +448,9 @@
       }
     },
     "node_modules/@esbuild/linux-riscv64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/linux-riscv64/-/linux-riscv64-0.28.0.tgz",
-      "integrity": "sha512-SQPZOwoTTT/HXFXQJG/vBX8sOFagGqvZyXcgLA3NhIqcBv1BJU1d46c0rGcrij2B56Z2rNiSLaZOYW5cUk7yLQ==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/linux-riscv64/-/linux-riscv64-0.28.1.tgz",
+      "integrity": "sha512-kw0owk1o0GFETUJyW0jc0G4Yzs0BHZn0JDZ8JRT088vjJYX777BAs1fDGxAC+q831qOs2DTC96mNsG2opdfyyQ==",
       "cpu": [
         "riscv64"
       ],
@@ -464,9 +464,9 @@
       }
     },
     "node_modules/@esbuild/linux-s390x": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/linux-s390x/-/linux-s390x-0.28.0.tgz",
-      "integrity": "sha512-SCfR0HN8CEEjnYnySJTd2cw0k9OHB/YFzt5zgJEwa+wL/T/raGWYMBqwDNAC6dqFKmJYZoQBRfHjgwLHGSrn3Q==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/linux-s390x/-/linux-s390x-0.28.1.tgz",
+      "integrity": "sha512-/lAIjX8aYFRByhh6L5rYtPEDRqa9de/4V/juOXcta5frjvzXO4/sqEtyytse0g3zZFuWu5cDN0MkLz2qRDD2Ag==",
       "cpu": [
         "s390x"
       ],
@@ -480,9 +480,9 @@
       }
     },
     "node_modules/@esbuild/linux-x64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/linux-x64/-/linux-x64-0.28.0.tgz",
-      "integrity": "sha512-us0dSb9iFxIi8srnpl931Nvs65it/Jd2a2K3qs7fz2WfGPHqzfzZTfec7oxZJRNPXPnNYZtanmRc4AL/JwVzHQ==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/linux-x64/-/linux-x64-0.28.1.tgz",
+      "integrity": "sha512-u/anNYF2mmVOEDwLtnQ1wOr3EZ9sTNGLWrsYGYwHWzGA3Si84IOkHXlbWTD1NB+9/1lcnweYKO54uhxZydNzfA==",
       "cpu": [
         "x64"
       ],
@@ -496,9 +496,9 @@
       }
     },
     "node_modules/@esbuild/netbsd-arm64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/netbsd-arm64/-/netbsd-arm64-0.28.0.tgz",
-      "integrity": "sha512-CR/RYotgtCKwtftMwJlUU7xCVNg3lMYZ0RzTmAHSfLCXw3NtZtNpswLEj/Kkf6kEL3Gw+BpOekRX0BYCtklhUw==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/netbsd-arm64/-/netbsd-arm64-0.28.1.tgz",
+      "integrity": "sha512-oks0DYbLwWMmaakTsCb+zL4E+aHRVLom9IJZOAthMQEPiQmydXHkziYEsGYRx0uNV/IjEKGAV941JzH02pflqw==",
       "cpu": [
         "arm64"
       ],
@@ -512,9 +512,9 @@
       }
     },
     "node_modules/@esbuild/netbsd-x64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/netbsd-x64/-/netbsd-x64-0.28.0.tgz",
-      "integrity": "sha512-nU1yhmYutL+fQ71Kxnhg8uEOdC0pwEW9entHykTgEbna2pw2dkbFSMeqjjyHZoCmt8SBkOSvV+yNmm94aUrrqw==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/netbsd-x64/-/netbsd-x64-0.28.1.tgz",
+      "integrity": "sha512-aeL6lAnN89Hz43Mlh1G8ARasbuoYvSITDEx0tHh5b7jJnHcssqgjy9Yx430GDpmCa6OyrKoS0aNRjKundRizGg==",
       "cpu": [
         "x64"
       ],
@@ -528,9 +528,9 @@
       }
     },
     "node_modules/@esbuild/openbsd-arm64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/openbsd-arm64/-/openbsd-arm64-0.28.0.tgz",
-      "integrity": "sha512-cXb5vApOsRsxsEl4mcZ1XY3D4DzcoMxR/nnc4IyqYs0rTI8ZKmW6kyyg+11Z8yvgMfAEldKzP7AdP64HnSC/6g==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/openbsd-arm64/-/openbsd-arm64-0.28.1.tgz",
+      "integrity": "sha512-MEFJe5C3R8pwXdZ5Y21oo6m7ePiS0d9pWucn99O/wvyJZChoIQKrQDxKrGeW8F5+T0okTHesAmDeiHDTIq0V/Q==",
       "cpu": [
         "arm64"
       ],
@@ -544,9 +544,9 @@
       }
     },
     "node_modules/@esbuild/openbsd-x64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/openbsd-x64/-/openbsd-x64-0.28.0.tgz",
-      "integrity": "sha512-8wZM2qqtv9UP3mzy7HiGYNH/zjTA355mpeuA+859TyR+e+Tc08IHYpLJuMsfpDJwoLo1ikIJI8jC3GFjnRClzA==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/openbsd-x64/-/openbsd-x64-0.28.1.tgz",
+      "integrity": "sha512-i/ZLIOafE0Z8cI/XANJAixoJL/uRAoS2xOA3rb0xN+KK0K177cMAsQYkzHtBrtMXAKuAc7HGgcWiZ/sRC1Nxgw==",
       "cpu": [
         "x64"
       ],
@@ -560,9 +560,9 @@
       }
     },
     "node_modules/@esbuild/openharmony-arm64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/openharmony-arm64/-/openharmony-arm64-0.28.0.tgz",
-      "integrity": "sha512-FLGfyizszcef5C3YtoyQDACyg95+dndv79i2EekILBofh5wpCa1KuBqOWKrEHZg3zrL3t5ouE5jgr94vA+Wb2w==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/openharmony-arm64/-/openharmony-arm64-0.28.1.tgz",
+      "integrity": "sha512-ge+Z7EXFNt2BO1oAMsVpiQ8EwndV9i1xXerAeTIK7AtPs3bKFXQM7nlRxDSIUIMeueR1CNXxqztLzdNeReKBJg==",
       "cpu": [
         "arm64"
       ],
@@ -576,9 +576,9 @@
       }
     },
     "node_modules/@esbuild/sunos-x64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/sunos-x64/-/sunos-x64-0.28.0.tgz",
-      "integrity": "sha512-1ZgjUoEdHZZl/YlV76TSCz9Hqj9h9YmMGAgAPYd+q4SicWNX3G5GCyx9uhQWSLcbvPW8Ni7lj4gDa1T40akdlw==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/sunos-x64/-/sunos-x64-0.28.1.tgz",
+      "integrity": "sha512-BEjgtECkL3vY+SaSQ6nzVfiALUeFxpawyp8Jmf5PtYhf1Ug40N1h/hxlhts+f1FvSvarEigdxS3BlSMI2PJLcQ==",
       "cpu": [
         "x64"
       ],
@@ -592,9 +592,9 @@
       }
     },
     "node_modules/@esbuild/win32-arm64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/win32-arm64/-/win32-arm64-0.28.0.tgz",
-      "integrity": "sha512-Q9StnDmQ/enxnpxCCLSg0oo4+34B9TdXpuyPeTedN/6+iXBJ4J+zwfQI28u/Jl40nOYAxGoNi7mFP40RUtkmUA==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/win32-arm64/-/win32-arm64-0.28.1.tgz",
+      "integrity": "sha512-lCv9eK/H6ZJWbE7bh2nw54CZ9M2nupBxJcTsdk/QQnWkdSjKGuxmmH8/GWrlT1eMmZfn4dGcCjRte397WqfQXA==",
       "cpu": [
         "arm64"
       ],
@@ -608,9 +608,9 @@
       }
     },
     "node_modules/@esbuild/win32-ia32": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/win32-ia32/-/win32-ia32-0.28.0.tgz",
-      "integrity": "sha512-zF3ag/gfiCe6U2iczcRzSYJKH1DCI+ByzSENHlM2FcDbEeo5Zd2C86Aq0tKUYAJJ1obRP84ymxIAksZUcdztHA==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/win32-ia32/-/win32-ia32-0.28.1.tgz",
+      "integrity": "sha512-zvb/mB2bSCoJOpoCBgYKKpX6YM6mJBlBUVUtVj41DlZJVEB6/0CKlRYxP5wWl1C1ILiCoAU5wZZ4q1P3qeS6Eg==",
       "cpu": [
         "ia32"
       ],
@@ -624,9 +624,9 @@
       }
     },
     "node_modules/@esbuild/win32-x64": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/@esbuild/win32-x64/-/win32-x64-0.28.0.tgz",
-      "integrity": "sha512-pEl1bO9mfAmIC+tW5btTmrKaujg3zGtUmWNdCw/xs70FBjwAL3o9OEKNHvNmnyylD6ubxUERiEhdsL0xBQ9efw==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/@esbuild/win32-x64/-/win32-x64-0.28.1.tgz",
+      "integrity": "sha512-bm4Mowrv+GXMlpWX++EcXw/iLyd1o3+bJkC2DkWXYVvgZCqD/bSj9ctZeAMC3cIxgjRVR2Dufaiu4YPxr5gW1A==",
       "cpu": [
         "x64"
       ],
@@ -1208,9 +1208,9 @@
       }
     },
     "node_modules/esbuild": {
-      "version": "0.28.0",
-      "resolved": "https://registry.npmjs.org/esbuild/-/esbuild-0.28.0.tgz",
-      "integrity": "sha512-sNR9MHpXSUV/XB4zmsFKN+QgVG82Cc7+/aaxJ8Adi8hyOac+EXptIp45QBPaVyX3N70664wRbTcLTOemCAnyqw==",
+      "version": "0.28.1",
+      "resolved": "https://registry.npmjs.org/esbuild/-/esbuild-0.28.1.tgz",
+      "integrity": "sha512-HrJrvZv5ayxBzPfwphOoNzkzOIIlifzk0KJrGK2c8R4+LKpMtpYLQeUdjnwjWv/LZlkH2laZk+4w78pi99D4Vw==",
       "hasInstallScript": true,
       "license": "MIT",
       "bin": {
@@ -1220,32 +1220,32 @@
         "node": ">=18"
       },
       "optionalDependencies": {
-        "@esbuild/aix-ppc64": "0.28.0",
-        "@esbuild/android-arm": "0.28.0",
-        "@esbuild/android-arm64": "0.28.0",
-        "@esbuild/android-x64": "0.28.0",
-        "@esbuild/darwin-arm64": "0.28.0",
-        "@esbuild/darwin-x64": "0.28.0",
-        "@esbuild/freebsd-arm64": "0.28.0",
-        "@esbuild/freebsd-x64": "0.28.0",
-        "@esbuild/linux-arm": "0.28.0",
-        "@esbuild/linux-arm64": "0.28.0",
-        "@esbuild/linux-ia32": "0.28.0",
-        "@esbuild/linux-loong64": "0.28.0",
-        "@esbuild/linux-mips64el": "0.28.0",
-        "@esbuild/linux-ppc64": "0.28.0",
-        "@esbuild/linux-riscv64": "0.28.0",
-        "@esbuild/linux-s390x": "0.28.0",
-        "@esbuild/linux-x64": "0.28.0",
-        "@esbuild/netbsd-arm64": "0.28.0",
-        "@esbuild/netbsd-x64": "0.28.0",
-        "@esbuild/openbsd-arm64": "0.28.0",
-        "@esbuild/openbsd-x64": "0.28.0",
-        "@esbuild/openharmony-arm64": "0.28.0",
-        "@esbuild/sunos-x64": "0.28.0",
-        "@esbuild/win32-arm64": "0.28.0",
-        "@esbuild/win32-ia32": "0.28.0",
-        "@esbuild/win32-x64": "0.28.0"
+        "@esbuild/aix-ppc64": "0.28.1",
+        "@esbuild/android-arm": "0.28.1",
+        "@esbuild/android-arm64": "0.28.1",
+        "@esbuild/android-x64": "0.28.1",
+        "@esbuild/darwin-arm64": "0.28.1",
+        "@esbuild/darwin-x64": "0.28.1",
+        "@esbuild/freebsd-arm64": "0.28.1",
+        "@esbuild/freebsd-x64": "0.28.1",
+        "@esbuild/linux-arm": "0.28.1",
+        "@esbuild/linux-arm64": "0.28.1",
+        "@esbuild/linux-ia32": "0.28.1",
+        "@esbuild/linux-loong64": "0.28.1",
+        "@esbuild/linux-mips64el": "0.28.1",
+        "@esbuild/linux-ppc64": "0.28.1",
+        "@esbuild/linux-riscv64": "0.28.1",
+        "@esbuild/linux-s390x": "0.28.1",
+        "@esbuild/linux-x64": "0.28.1",
+        "@esbuild/netbsd-arm64": "0.28.1",
+        "@esbuild/netbsd-x64": "0.28.1",
+        "@esbuild/openbsd-arm64": "0.28.1",
+        "@esbuild/openbsd-x64": "0.28.1",
+        "@esbuild/openharmony-arm64": "0.28.1",
+        "@esbuild/sunos-x64": "0.28.1",
+        "@esbuild/win32-arm64": "0.28.1",
+        "@esbuild/win32-ia32": "0.28.1",
+        "@esbuild/win32-x64": "0.28.1"
       }
     },
     "node_modules/escape-html": {

extensions/anthropic-vertex/stream-runtime.test.ts

@@ -3,8 +3,6 @@ import { createAssistantMessageEventStream, type Model } from "openclaw/plugin-s
 import { beforeAll, describe, expect, it, vi } from "vitest";
 import type { AnthropicVertexStreamDeps } from "./stream-runtime.js";
 
-const SYSTEM_PROMPT_CACHE_BOUNDARY = "\n<!-- OPENCLAW_CACHE_BOUNDARY -->\n";
-
 function createStreamDeps(): {
   deps: AnthropicVertexStreamDeps;
   streamAnthropicMock: ReturnType<typeof vi.fn>;
@@ -50,8 +48,6 @@ function makeModel(params: {
   } as Model<"anthropic-messages">;
 }
 
-const CACHE_BOUNDARY_PROMPT = `Stable prefix${SYSTEM_PROMPT_CACHE_BOUNDARY}Dynamic suffix`;
-
 type PayloadHook = (payload: unknown, payloadModel: unknown) => Promise<unknown>;
 
 function streamAnthropicCall(streamAnthropicMock: ReturnType<typeof vi.fn>): unknown[] {
@@ -72,8 +68,8 @@ function streamTransportOptions(
   return options as Record<string, unknown>;
 }
 
-function captureCacheBoundaryPayloadHook(
-  onPayload: PayloadHook,
+function captureTransportPayloadHook(
+  onPayload: PayloadHook | undefined,
   deps: AnthropicVertexStreamDeps,
   streamAnthropicMock: ReturnType<typeof vi.fn>,
 ) {
@@ -82,14 +78,8 @@ function captureCacheBoundaryPayloadHook(
 
   void streamFn(
     model,
-    {
-      systemPrompt: CACHE_BOUNDARY_PROMPT,
-      messages: [{ role: "user", content: "Hello" }],
-    } as never,
-    {
-      cacheRetention: "short",
-      onPayload,
-    } as never,
+    { messages: [{ role: "user", content: "Hello" }] } as never,
+    { cacheRetention: "short", ...(onPayload ? { onPayload } : {}) } as never,
   );
 
   const transportOptions = streamTransportOptions(streamAnthropicMock);
@@ -97,26 +87,30 @@ function captureCacheBoundaryPayloadHook(
   return { model, onPayload: transportOptions.onPayload as PayloadHook | undefined };
 }
 
-function buildExpectedCacheBoundaryPayload(messageText: string) {
+// Mirrors the shared anthropic-messages transport output: cache boundary already
+// split (uncached dynamic suffix) and all four cache_control markers allocated.
+function buildBudgetedTransportPayload() {
   return {
     system: [
-      {
-        type: "text",
-        text: "Stable prefix",
-        cache_control: { type: "ephemeral" },
-      },
-      {
-        type: "text",
-        text: "Dynamic suffix",
-      },
+      { type: "text", text: "Stable prefix", cache_control: { type: "ephemeral" } },
+      { type: "text", text: "Dynamic suffix" },
+    ],
+    tools: [
+      { name: "exec", input_schema: { type: "object" }, cache_control: { type: "ephemeral" } },
     ],
     messages: [
+      {
+        role: "user",
+        content: [{ type: "text", text: "Hello", cache_control: { type: "ephemeral" } }],
+      },
+      { role: "assistant", content: [{ type: "tool_use", id: "t1", name: "exec", input: {} }] },
       {
         role: "user",
         content: [
           {
-            type: "text",
-            text: messageText,
+            type: "tool_result",
+            tool_use_id: "t1",
+            content: [],
             cache_control: { type: "ephemeral" },
           },
         ],
@@ -125,6 +119,29 @@ function buildExpectedCacheBoundaryPayload(messageText: string) {
   };
 }
 
+function countCacheControlMarkers(payload: unknown): number {
+  let count = 0;
+  const visit = (value: unknown) => {
+    if (Array.isArray(value)) {
+      value.forEach(visit);
+      return;
+    }
+    if (!value || typeof value !== "object") {
+      return;
+    }
+    const record = value as Record<string, unknown>;
+    if (record.cache_control !== undefined) {
+      count += 1;
+    }
+    visit(record.content);
+  };
+  const record = payload as Record<string, unknown>;
+  visit(record.system);
+  visit(record.tools);
+  visit(record.messages);
+  return count;
+}
+
 describe("createAnthropicVertexStreamFn", () => {
   beforeAll(async () => {
     ({ createAnthropicVertexStreamFn, createAnthropicVertexStreamFnForModel } =
@@ -343,63 +360,35 @@ describe("createAnthropicVertexStreamFn", () => {
     expect(transportOptions).not.toHaveProperty("temperature");
   });
 
-  it("applies Anthropic cache-boundary shaping before forwarding payload hooks", async () => {
+  it("keeps already-budgeted cache_control markers intact when forwarding payload hooks", async () => {
     const { deps, streamAnthropicMock } = createStreamDeps();
     const onPayload = vi.fn(async (payload: unknown) => payload);
-    const { model, onPayload: transportPayloadHook } = captureCacheBoundaryPayloadHook(
+    const { model, onPayload: transportPayloadHook } = captureTransportPayloadHook(
       onPayload,
       deps,
       streamAnthropicMock,
     );
-    const payload = {
-      system: [
-        {
-          type: "text",
-          text: CACHE_BOUNDARY_PROMPT,
-          cache_control: { type: "ephemeral" },
-        },
-      ],
-      messages: [{ role: "user", content: "Hello" }],
-    };
+    const payload = buildBudgetedTransportPayload();
 
     const nextPayload = await transportPayloadHook?.(payload, model);
 
-    const expectedPayload = buildExpectedCacheBoundaryPayload("Hello");
-    expect(onPayload).toHaveBeenCalledWith(expectedPayload, model);
-    expect(nextPayload).toEqual(expectedPayload);
+    expect(onPayload).toHaveBeenCalledWith(payload, model);
+    expect(countCacheControlMarkers(nextPayload)).toBe(4);
+    expect((nextPayload as ReturnType<typeof buildBudgetedTransportPayload>).system[1]).toEqual({
+      type: "text",
+      text: "Dynamic suffix",
+    });
   });
 
-  it("reapplies Anthropic cache-boundary shaping when payload hooks return a fresh payload", async () => {
+  it("omits the transport payload hook when the caller provides none", () => {
     const { deps, streamAnthropicMock } = createStreamDeps();
-    const onPayload = vi.fn(async () => ({
-      system: [
-        {
-          type: "text",
-          text: CACHE_BOUNDARY_PROMPT,
-        },
-      ],
-      messages: [{ role: "user", content: "Hello again" }],
-    }));
-    const { model, onPayload: transportPayloadHook } = captureCacheBoundaryPayloadHook(
-      onPayload,
+    const { onPayload: transportPayloadHook } = captureTransportPayloadHook(
+      undefined,
       deps,
       streamAnthropicMock,
     );
 
-    const nextPayload = await transportPayloadHook?.(
-      {
-        system: [
-          {
-            type: "text",
-            text: CACHE_BOUNDARY_PROMPT,
-          },
-        ],
-        messages: [{ role: "user", content: "Hello" }],
-      },
-      model,
-    );
-
-    expect(nextPayload).toEqual(buildExpectedCacheBoundaryPayload("Hello again"));
+    expect(transportPayloadHook).toBeUndefined();
   });
 
   it("omits maxTokens when neither the model nor request provide a finite limit", () => {

extensions/anthropic-vertex/stream-runtime.ts

@@ -1,6 +1,6 @@
 /**
  * Anthropic Vertex stream runtime. It constructs Vertex SDK clients and adapts
- * OpenClaw stream options into Anthropic Messages payload policy.
+ * OpenClaw stream options for the shared Anthropic Messages transport.
  */
 import { AnthropicVertex as AnthropicVertexSdk } from "@anthropic-ai/vertex-sdk";
 import type { StreamFn } from "openclaw/plugin-sdk/agent-core";
@@ -18,10 +18,6 @@ import {
   supportsClaudeNativeMaxEffort,
   supportsClaudeNativeXhighEffort,
 } from "openclaw/plugin-sdk/provider-model-shared";
-import {
-  applyAnthropicPayloadPolicyToParams,
-  resolveAnthropicPayloadPolicy,
-} from "openclaw/plugin-sdk/provider-stream-shared";
 import { resolveAnthropicVertexClientRegion, resolveAnthropicVertexProjectId } from "./region.js";
 
 type AnthropicVertexTransportOptions = ProviderStreamOptions & {
@@ -120,36 +116,6 @@ function resolveAnthropicVertexMaxTokens(params: {
   return requested ?? modelMax;
 }
 
-function createAnthropicVertexOnPayload(params: {
-  model: { api: string; baseUrl?: string; provider: string };
-  cacheRetention: ProviderStreamOptions["cacheRetention"] | undefined;
-  onPayload: ProviderStreamOptions["onPayload"] | undefined;
-}): NonNullable<ProviderStreamOptions["onPayload"]> {
-  const policy = resolveAnthropicPayloadPolicy({
-    provider: params.model.provider,
-    api: params.model.api,
-    baseUrl: params.model.baseUrl,
-    cacheRetention: params.cacheRetention,
-    enableCacheControl: true,
-  });
-
-  function applyPolicy(payload: unknown): unknown {
-    if (payload && typeof payload === "object" && !Array.isArray(payload)) {
-      applyAnthropicPayloadPolicyToParams(payload as Record<string, unknown>, policy);
-    }
-    return payload;
-  }
-
-  return async (payload, model) => {
-    const shapedPayload = applyPolicy(payload);
-    const nextPayload = await params.onPayload?.(shapedPayload, model);
-    if (nextPayload === undefined || nextPayload === shapedPayload) {
-      return shapedPayload;
-    }
-    return applyPolicy(nextPayload);
-  };
-}
-
 /**
  * Create a StreamFn that routes through OpenClaw's generic model stream with an
  * injected `AnthropicVertex` client.  All streaming, message conversion, and
@@ -200,11 +166,10 @@ export function createAnthropicVertexStreamFn(
       cacheRetention: options?.cacheRetention,
       sessionId: options?.sessionId,
       headers: options?.headers,
-      onPayload: createAnthropicVertexOnPayload({
-        model: transportModel,
-        cacheRetention: options?.cacheRetention,
-        onPayload: options?.onPayload,
-      }),
+      // The shared anthropic-messages transport already splits the system prompt
+      // cache boundary and budgets all cache_control markers; re-applying the
+      // payload policy here marked the uncached suffix and breached the 4-marker cap.
+      onPayload: options?.onPayload,
       maxRetryDelayMs: options?.maxRetryDelayMs,
       metadata: options?.metadata,
     };

extensions/browser/src/cli/browser-cli-manage.test.ts

@@ -461,4 +461,24 @@ describe("browser manage output", () => {
     expect(output).toContain("OK gateway: browser control endpoint reachable");
     expect(output).toContain("OK tabs: 1 visible, use tab reference t1");
   });
+
+  it("prints a readable browser doctor failure when gateway auth SecretRefs are unavailable", async () => {
+    const error = Object.assign(new Error("gateway.auth.password unavailable"), {
+      code: "GATEWAY_SECRET_REF_UNAVAILABLE",
+      name: "GatewaySecretRefUnavailableError",
+    });
+    getBrowserManageCallBrowserRequestMock().mockRejectedValueOnce(error);
+
+    const program = createBrowserManageProgram();
+    await expect(program.parseAsync(["browser", "doctor"], { from: "user" })).rejects.toThrow(
+      "__exit__:1",
+    );
+
+    const output = lastRuntimeLog();
+    expect(output).toContain(
+      "FAIL gateway: Gateway auth SecretRef is unavailable in this command path",
+    );
+    expect(output).toContain("OPENCLAW_GATEWAY_TOKEN");
+    expect(output).not.toContain("GatewaySecretRefUnavailableError");
+  });
 });

extensions/browser/src/cli/browser-cli-manage.ts

@@ -152,6 +152,24 @@ function formatDoctorLine(check: BrowserDoctorCheck): string {
   return `${check.ok ? "OK" : "FAIL"} ${check.name}${check.detail ? `: ${check.detail}` : ""}`;
 }
 
+function isGatewaySecretRefUnavailableErrorShape(error: unknown): boolean {
+  if (!(error instanceof Error)) {
+    return false;
+  }
+  const errorRecord = error as Error & { code?: unknown };
+  return (
+    errorRecord.name === "GatewaySecretRefUnavailableError" ||
+    errorRecord.code === "GATEWAY_SECRET_REF_UNAVAILABLE"
+  );
+}
+
+function formatBrowserDoctorGatewayError(error: unknown): string {
+  if (!isGatewaySecretRefUnavailableErrorShape(error)) {
+    return String(error);
+  }
+  return "Gateway auth SecretRef is unavailable in this command path; browser doctor cannot reach the admin-scoped browser.request endpoint. Set OPENCLAW_GATEWAY_TOKEN or OPENCLAW_GATEWAY_PASSWORD, then retry.";
+}
+
 async function runBrowserDoctor(parent: BrowserParentOpts, profile?: string, deep?: boolean) {
   const checks: BrowserDoctorCheck[] = [];
   let status: BrowserStatus | null;
@@ -167,7 +185,7 @@ async function runBrowserDoctor(parent: BrowserParentOpts, profile?: string, dee
     checks.push({
       name: "gateway",
       ok: false,
-      detail: String(err),
+      detail: formatBrowserDoctorGatewayError(err),
     });
     return { ok: false, checks };
   }

extensions/codex/openclaw.plugin.json

@@ -100,7 +100,7 @@
             "default": false
           },
           "allow_destructive_actions": {
-            "oneOf": [{ "type": "boolean" }, { "const": "on-request" }],
+            "type": "boolean",
             "default": true
           },
           "plugins": {
@@ -120,7 +120,7 @@
                   "type": "string"
                 },
                 "allow_destructive_actions": {
-                  "oneOf": [{ "type": "boolean" }, { "const": "on-request" }]
+                  "type": "boolean"
                 }
               }
             }
@@ -290,7 +290,7 @@
     },
     "codexPlugins.allow_destructive_actions": {
       "label": "Allow Destructive Plugin Actions",
-      "help": "Default policy for plugin app write or destructive action elicitations. Use true to auto-accept safe schemas, false to decline, or on-request to ask through plugin approvals.",
+      "help": "Default policy for plugin app write or destructive action elicitations. Defaults to true.",
       "advanced": true
     },
     "codexPlugins.plugins": {

extensions/codex/src/app-server/attempt-context.ts

@@ -15,6 +15,7 @@ import {
   type EmbeddedRunAttemptResult,
 } from "openclaw/plugin-sdk/agent-harness-runtime";
 import { resolveAgentWorkspaceDir } from "openclaw/plugin-sdk/agent-runtime";
+import { buildMemorySystemPromptAddition } from "openclaw/plugin-sdk/core";
 import type { CodexDynamicToolSpec, JsonValue } from "./protocol.js";
 import { isJsonObject } from "./protocol.js";
 import type { CodexAppServerThreadBinding } from "./session-binding.js";
@@ -249,9 +250,11 @@ export async function buildCodexWorkspaceBootstrapContext(params: {
         turnScopedDeveloperInstructionFiles,
       ),
       memoryCollaborationInstructions: shouldInjectCodexOpenClawPromptContext(params.params)
-        ? renderCodexWorkspaceMemoryReference({
+        ? renderCodexWorkspaceMemoryCollaborationInstructions({
             files: memoryReferenceFiles,
             toolNames: params.memoryToolNames,
+            memoryToolRouted: memoryToolsAvailable,
+            citationsMode: params.params.config?.memory?.citations,
           })
         : undefined,
       heartbeatCollaborationInstructions:
@@ -805,6 +808,55 @@ export function renderCodexWorkspaceMemoryReference(params: {
   return lines.join("\n").trim();
 }
 
+function renderCodexWorkspaceMemoryCollaborationInstructions(params: {
+  files: EmbeddedContextFile[];
+  toolNames: readonly string[];
+  memoryToolRouted: boolean;
+  citationsMode?: Parameters<typeof buildMemorySystemPromptAddition>[0]["citationsMode"];
+}): string | undefined {
+  const memoryRecallInstructions = params.memoryToolRouted
+    ? renderCodexMemoryRecallInstructions({
+        toolNames: params.toolNames,
+        citationsMode: params.citationsMode,
+      })
+    : undefined;
+  const memoryReferenceInstructions = renderCodexWorkspaceMemoryReference({
+    files: params.files,
+    toolNames: params.toolNames,
+  });
+  const sections = [memoryRecallInstructions, memoryReferenceInstructions].filter(isNonEmptyString);
+  return sections.length > 0 ? sections.join("\n\n") : undefined;
+}
+
+function renderCodexMemoryRecallInstructions(params: {
+  toolNames: readonly string[];
+  citationsMode?: Parameters<typeof buildMemorySystemPromptAddition>[0]["citationsMode"];
+}): string | undefined {
+  const availableTools = new Set(params.toolNames);
+  const memoryPrompt = buildMemorySystemPromptAddition({
+    availableTools,
+    citationsMode: params.citationsMode,
+  });
+  if (!memoryPrompt) {
+    // Memory recall policy belongs to the active memory plugin.
+    // Codex-side fallback text can mask plugin lifecycle bugs or misdescribe third-party memory tools.
+    return undefined;
+  }
+  const toolSearchBridge = renderCodexMemoryToolSearchBridge(params.toolNames);
+  return [memoryPrompt, toolSearchBridge].filter(isNonEmptyString).join("\n").trim();
+}
+
+function renderCodexMemoryToolSearchBridge(toolNames: readonly string[]): string | undefined {
+  const memoryToolNames = toolNames
+    .map((name) => normalizeCodexDynamicToolName(name))
+    .filter((name) => CODEX_MEMORY_TOOL_NAMES.has(name))
+    .toSorted();
+  if (memoryToolNames.length === 0) {
+    return undefined;
+  }
+  return `Codex may expose ${memoryToolNames.join(" and ")} as deferred tools. When the memory guidance above calls for memory recall, use an already-loaded memory tool directly. If the needed memory tool is deferred and not currently callable, use \`tool_search\` to load it, then call that memory tool.`;
+}
+
 /** Returns whether the current dynamic tool list can serve workspace memory. */
 export function hasCodexWorkspaceMemoryTools(tools: readonly { name: string }[]): boolean {
   return getCodexWorkspaceMemoryToolNames(tools).length > 0;

extensions/codex/src/app-server/config.test.ts

@@ -859,7 +859,6 @@ allowed_sandbox_modes = ["read-only", "workspace-write"]
       configured: true,
       enabled: true,
       allowDestructiveActions: false,
-      destructiveApprovalMode: "deny",
       pluginPolicies: [
         {
           configKey: "google-calendar",
@@ -867,7 +866,6 @@ allowed_sandbox_modes = ["read-only", "workspace-write"]
           pluginName: "google-calendar",
           enabled: true,
           allowDestructiveActions: true,
-          destructiveApprovalMode: "auto",
         },
         {
           configKey: "slack",
@@ -875,88 +873,11 @@ allowed_sandbox_modes = ["read-only", "workspace-write"]
           pluginName: "slack",
           enabled: false,
           allowDestructiveActions: false,
-          destructiveApprovalMode: "deny",
         },
       ],
     });
   });
 
-  it("parses on-request native Codex plugin destructive policy", () => {
-    const config = readCodexPluginConfig({
-      codexPlugins: {
-        enabled: true,
-        allow_destructive_actions: "on-request",
-        plugins: {
-          "google-calendar": {
-            marketplaceName: "openai-curated",
-            pluginName: "google-calendar",
-          },
-          slack: {
-            marketplaceName: "openai-curated",
-            pluginName: "slack",
-            allow_destructive_actions: false,
-          },
-          gmail: {
-            marketplaceName: "openai-curated",
-            pluginName: "gmail",
-            allow_destructive_actions: true,
-          },
-        },
-      },
-    });
-
-    expect(config.codexPlugins?.allow_destructive_actions).toBe("on-request");
-    expect(resolveCodexPluginsPolicy(config)).toEqual({
-      configured: true,
-      enabled: true,
-      allowDestructiveActions: true,
-      destructiveApprovalMode: "on-request",
-      pluginPolicies: [
-        {
-          configKey: "gmail",
-          marketplaceName: "openai-curated",
-          pluginName: "gmail",
-          enabled: true,
-          allowDestructiveActions: true,
-          destructiveApprovalMode: "auto",
-        },
-        {
-          configKey: "google-calendar",
-          marketplaceName: "openai-curated",
-          pluginName: "google-calendar",
-          enabled: true,
-          allowDestructiveActions: true,
-          destructiveApprovalMode: "on-request",
-        },
-        {
-          configKey: "slack",
-          marketplaceName: "openai-curated",
-          pluginName: "slack",
-          enabled: true,
-          allowDestructiveActions: false,
-          destructiveApprovalMode: "deny",
-        },
-      ],
-    });
-  });
-
-  it("rejects unsupported native Codex plugin destructive policy strings", () => {
-    const config = readCodexPluginConfig({
-      codexPlugins: {
-        enabled: true,
-        allow_destructive_actions: "ask",
-        plugins: {
-          slack: {
-            marketplaceName: "openai-curated",
-            pluginName: "slack",
-          },
-        },
-      },
-    });
-
-    expect(config.codexPlugins).toBeUndefined();
-  });
-
   it("defaults native Codex plugin destructive policy to enabled", () => {
     const policy = resolveCodexPluginsPolicy({
       codexPlugins: {
@@ -974,7 +895,6 @@ allowed_sandbox_modes = ["read-only", "workspace-write"]
       configured: true,
       enabled: true,
       allowDestructiveActions: true,
-      destructiveApprovalMode: "auto",
       pluginPolicies: [
         {
           configKey: "slack",
@@ -982,7 +902,6 @@ allowed_sandbox_modes = ["read-only", "workspace-write"]
           pluginName: "slack",
           enabled: true,
           allowDestructiveActions: true,
-          destructiveApprovalMode: "auto",
         },
       ],
     });

extensions/codex/src/app-server/config.ts

@@ -67,8 +67,7 @@ export type CodexAppServerSandboxMode = "read-only" | "workspace-write" | "dange
 type CodexAppServerApprovalsReviewer = "user" | "auto_review" | "guardian_subagent";
 type CodexAppServerCommandSource = "managed" | "resolved-managed" | "config" | "env";
 export type CodexDynamicToolsLoading = "searchable" | "direct";
-export type CodexPluginDestructivePolicy = boolean | "on-request";
-export type CodexPluginDestructiveApprovalMode = "auto" | "deny" | "on-request";
+export type CodexPluginDestructivePolicy = boolean;
 
 export const CODEX_PLUGINS_MARKETPLACE_NAME = "openai-curated";
 
@@ -116,15 +115,13 @@ export type ResolvedCodexPluginPolicy = {
   marketplaceName: typeof CODEX_PLUGINS_MARKETPLACE_NAME;
   pluginName: string;
   enabled: boolean;
-  allowDestructiveActions: boolean;
-  destructiveApprovalMode: CodexPluginDestructiveApprovalMode;
+  allowDestructiveActions: CodexPluginDestructivePolicy;
 };
 
 export type ResolvedCodexPluginsPolicy = {
   configured: boolean;
   enabled: boolean;
-  allowDestructiveActions: boolean;
-  destructiveApprovalMode: CodexPluginDestructiveApprovalMode;
+  allowDestructiveActions: CodexPluginDestructivePolicy;
   pluginPolicies: ResolvedCodexPluginPolicy[];
 };
 
@@ -261,7 +258,6 @@ const codexAppServerApprovalPolicySchema = z.enum([
 const codexAppServerSandboxSchema = z.enum(["read-only", "workspace-write", "danger-full-access"]);
 const codexAppServerApprovalsReviewerSchema = z.enum(["user", "auto_review", "guardian_subagent"]);
 const codexDynamicToolsLoadingSchema = z.enum(["searchable", "direct"]);
-const codexPluginDestructivePolicySchema = z.union([z.boolean(), z.literal("on-request")]);
 const codexAppServerServiceTierSchema = z
   .preprocess(
     (value) => (value === null ? null : normalizeCodexServiceTier(value)),
@@ -279,14 +275,14 @@ const codexPluginEntryConfigSchema = z
     enabled: z.boolean().optional(),
     marketplaceName: z.literal(CODEX_PLUGINS_MARKETPLACE_NAME).optional(),
     pluginName: z.string().trim().min(1).optional(),
-    allow_destructive_actions: codexPluginDestructivePolicySchema.optional(),
+    allow_destructive_actions: z.boolean().optional(),
   })
   .strict();
 
 const codexPluginsConfigSchema = z
   .object({
     enabled: z.boolean().optional(),
-    allow_destructive_actions: codexPluginDestructivePolicySchema.optional(),
+    allow_destructive_actions: z.boolean().optional(),
     plugins: z.record(z.string(), codexPluginEntryConfigSchema).optional(),
   })
   .strict();
@@ -384,25 +380,19 @@ export function resolveCodexPluginsPolicy(pluginConfig?: unknown): ResolvedCodex
   const config = readCodexPluginConfig(pluginConfig).codexPlugins;
   const configured = config !== undefined;
   const enabled = config?.enabled === true;
-  const destructivePolicy = resolveCodexPluginDestructivePolicy(
-    config?.allow_destructive_actions ?? true,
-  );
+  const allowDestructiveActions = config?.allow_destructive_actions ?? true;
   const pluginPolicies = Object.entries(config?.plugins ?? {})
     .flatMap(([configKey, entry]): ResolvedCodexPluginPolicy[] => {
       if (entry.marketplaceName !== CODEX_PLUGINS_MARKETPLACE_NAME || !entry.pluginName) {
         return [];
       }
-      const entryDestructivePolicy = resolveCodexPluginDestructivePolicy(
-        entry.allow_destructive_actions ?? config?.allow_destructive_actions ?? true,
-      );
       return [
         {
           configKey,
           marketplaceName: CODEX_PLUGINS_MARKETPLACE_NAME,
           pluginName: entry.pluginName,
           enabled: enabled && entry.enabled !== false,
-          allowDestructiveActions: entryDestructivePolicy.allowDestructiveActions,
-          destructiveApprovalMode: entryDestructivePolicy.destructiveApprovalMode,
+          allowDestructiveActions: entry.allow_destructive_actions ?? allowDestructiveActions,
         },
       ];
     })
@@ -410,25 +400,11 @@ export function resolveCodexPluginsPolicy(pluginConfig?: unknown): ResolvedCodex
   return {
     configured,
     enabled,
-    allowDestructiveActions: destructivePolicy.allowDestructiveActions,
-    destructiveApprovalMode: destructivePolicy.destructiveApprovalMode,
+    allowDestructiveActions,
     pluginPolicies,
   };
 }
 
-function resolveCodexPluginDestructivePolicy(policy: CodexPluginDestructivePolicy): {
-  allowDestructiveActions: boolean;
-  destructiveApprovalMode: CodexPluginDestructiveApprovalMode;
-} {
-  if (policy === "on-request") {
-    return { allowDestructiveActions: true, destructiveApprovalMode: "on-request" };
-  }
-  return {
-    allowDestructiveActions: policy,
-    destructiveApprovalMode: policy ? "auto" : "deny",
-  };
-}
-
 export function resolveCodexAppServerRuntimeOptions(
   params: {
     pluginConfig?: unknown;

extensions/codex/src/app-server/elicitation-bridge.test.ts

@@ -157,7 +157,6 @@ function buildConnectorPluginApprovalElicitation(overrides: Record<string, unkno
 function createPluginAppPolicyContext(
   params: {
     allowDestructiveActions?: boolean;
-    destructiveApprovalMode?: "auto" | "deny" | "on-request";
     apps?: Array<{ appId: string; pluginName: string; mcpServerNames: string[] }>;
   } = {},
 ) {
@@ -178,9 +177,6 @@ function createPluginAppPolicyContext(
           marketplaceName: "openai-curated" as const,
           pluginName: app.pluginName,
           allowDestructiveActions: params.allowDestructiveActions ?? false,
-          ...(params.destructiveApprovalMode
-            ? { destructiveApprovalMode: params.destructiveApprovalMode }
-            : {}),
           mcpServerNames: app.mcpServerNames,
         },
       ]),
@@ -835,242 +831,6 @@ describe("Codex app-server elicitation bridge", () => {
     expect(mockCallGatewayTool).not.toHaveBeenCalled();
   });
 
-  it("routes on-request connector-id plugin app elicitations through plugin approvals", async () => {
-    mockCallGatewayTool
-      .mockResolvedValueOnce({ id: "plugin:approval-calendar", status: "accepted" })
-      .mockResolvedValueOnce({ id: "plugin:approval-calendar", decision: "allow-once" });
-
-    const result = await handleCodexAppServerElicitationRequest({
-      requestParams: buildConnectorPluginApprovalElicitation(),
-      paramsForRun: createParams(),
-      threadId: "thread-1",
-      turnId: "turn-1",
-      pluginAppPolicyContext: createPluginAppPolicyContext({
-        allowDestructiveActions: true,
-        destructiveApprovalMode: "on-request",
-        apps: [
-          {
-            appId: "connector_google_calendar",
-            pluginName: "google-calendar",
-            mcpServerNames: [],
-          },
-        ],
-      }),
-    });
-
-    expect(result).toEqual({
-      action: "accept",
-      content: null,
-      _meta: null,
-    });
-    expect(mockCallGatewayTool.mock.calls.map(([method]) => method)).toEqual([
-      "plugin.approval.request",
-      "plugin.approval.waitDecision",
-    ]);
-    expect(gatewayToolArg(0, 2)).toMatchObject({
-      allowedDecisions: ["allow-once", "deny"],
-      title: "Allow Google Calendar to create an event?",
-      toolName: "codex_mcp_tool_approval",
-      twoPhase: true,
-    });
-  });
-
-  it("maps on-request plugin allow-always only when Codex offers always persistence", async () => {
-    mockCallGatewayTool
-      .mockResolvedValueOnce({ id: "plugin:approval-calendar-always", status: "accepted" })
-      .mockResolvedValueOnce({
-        id: "plugin:approval-calendar-always",
-        decision: "allow-always",
-      });
-
-    const result = await handleCodexAppServerElicitationRequest({
-      requestParams: buildConnectorPluginApprovalElicitation({
-        _meta: {
-          codex_approval_kind: "mcp_tool_call",
-          source: "connector",
-          connector_id: "connector_google_calendar",
-          connector_name: "Google Calendar",
-          persist: ["session", "always"],
-          tool_title: "create_event",
-        },
-      }),
-      paramsForRun: createParams(),
-      threadId: "thread-1",
-      turnId: "turn-1",
-      pluginAppPolicyContext: createPluginAppPolicyContext({
-        allowDestructiveActions: true,
-        destructiveApprovalMode: "on-request",
-        apps: [
-          {
-            appId: "connector_google_calendar",
-            pluginName: "google-calendar",
-            mcpServerNames: [],
-          },
-        ],
-      }),
-    });
-
-    expect(result).toEqual({
-      action: "accept",
-      content: null,
-      _meta: {
-        persist: "always",
-      },
-    });
-    expect(gatewayToolArg(0, 2)).toMatchObject({
-      allowedDecisions: ["allow-once", "allow-always", "deny"],
-    });
-  });
-
-  it("does not expose allow-always for on-request plugin session-only persistence", async () => {
-    mockCallGatewayTool
-      .mockResolvedValueOnce({ id: "plugin:approval-calendar-session", status: "accepted" })
-      .mockResolvedValueOnce({
-        id: "plugin:approval-calendar-session",
-        decision: "allow-once",
-      });
-
-    const result = await handleCodexAppServerElicitationRequest({
-      requestParams: buildConnectorPluginApprovalElicitation({
-        _meta: {
-          codex_approval_kind: "mcp_tool_call",
-          source: "connector",
-          connector_id: "connector_google_calendar",
-          connector_name: "Google Calendar",
-          persist: ["session"],
-          tool_title: "create_event",
-        },
-        requestedSchema: {
-          type: "object",
-          properties: {
-            approve: {
-              type: "boolean",
-              title: "Approve this app action",
-            },
-            persist: {
-              type: "string",
-              title: "Persist choice",
-              enum: ["session", "always"],
-            },
-          },
-          required: ["approve"],
-        },
-      }),
-      paramsForRun: createParams(),
-      threadId: "thread-1",
-      turnId: "turn-1",
-      pluginAppPolicyContext: createPluginAppPolicyContext({
-        allowDestructiveActions: true,
-        destructiveApprovalMode: "on-request",
-        apps: [
-          {
-            appId: "connector_google_calendar",
-            pluginName: "google-calendar",
-            mcpServerNames: [],
-          },
-        ],
-      }),
-    });
-
-    expect(result).toEqual({
-      action: "accept",
-      content: {
-        approve: true,
-      },
-      _meta: null,
-    });
-    expect(gatewayToolArg(0, 2)).toMatchObject({
-      allowedDecisions: ["allow-once", "deny"],
-    });
-  });
-
-  it("declines denied on-request plugin app approvals", async () => {
-    mockCallGatewayTool
-      .mockResolvedValueOnce({ id: "plugin:approval-calendar-deny", status: "accepted" })
-      .mockResolvedValueOnce({ id: "plugin:approval-calendar-deny", decision: "deny" });
-
-    const result = await handleCodexAppServerElicitationRequest({
-      requestParams: buildConnectorPluginApprovalElicitation(),
-      paramsForRun: createParams(),
-      threadId: "thread-1",
-      turnId: "turn-1",
-      pluginAppPolicyContext: createPluginAppPolicyContext({
-        allowDestructiveActions: true,
-        destructiveApprovalMode: "on-request",
-        apps: [
-          {
-            appId: "connector_google_calendar",
-            pluginName: "google-calendar",
-            mcpServerNames: [],
-          },
-        ],
-      }),
-    });
-
-    expect(result).toEqual({ action: "decline", content: null, _meta: null });
-  });
-
-  it("fails closed when on-request plugin approval routing is unavailable", async () => {
-    mockCallGatewayTool.mockResolvedValueOnce({
-      id: "plugin:approval-calendar-unavailable",
-      decision: null,
-    });
-
-    const result = await handleCodexAppServerElicitationRequest({
-      requestParams: buildConnectorPluginApprovalElicitation(),
-      paramsForRun: createParams(),
-      threadId: "thread-1",
-      turnId: "turn-1",
-      pluginAppPolicyContext: createPluginAppPolicyContext({
-        allowDestructiveActions: true,
-        destructiveApprovalMode: "on-request",
-        apps: [
-          {
-            appId: "connector_google_calendar",
-            pluginName: "google-calendar",
-            mcpServerNames: [],
-          },
-        ],
-      }),
-    });
-
-    expect(result).toEqual({ action: "decline", content: null, _meta: null });
-    expect(mockCallGatewayTool.mock.calls.map(([method]) => method)).toEqual([
-      "plugin.approval.request",
-    ]);
-  });
-
-  it("cancels on-request plugin app approvals when the turn aborts", async () => {
-    const abortController = new AbortController();
-    mockCallGatewayTool
-      .mockResolvedValueOnce({ id: "plugin:approval-calendar-abort", status: "accepted" })
-      .mockImplementationOnce(() => {
-        abortController.abort(new Error("turn stopped"));
-        return new Promise(() => undefined);
-      });
-
-    const result = await handleCodexAppServerElicitationRequest({
-      requestParams: buildConnectorPluginApprovalElicitation(),
-      paramsForRun: createParams(),
-      threadId: "thread-1",
-      turnId: "turn-1",
-      pluginAppPolicyContext: createPluginAppPolicyContext({
-        allowDestructiveActions: true,
-        destructiveApprovalMode: "on-request",
-        apps: [
-          {
-            appId: "connector_google_calendar",
-            pluginName: "google-calendar",
-            mcpServerNames: [],
-          },
-        ],
-      }),
-      signal: abortController.signal,
-    });
-
-    expect(result).toEqual({ action: "cancel", content: null, _meta: null });
-  });
-
   it("declines connector-id plugin app elicitations when destructive actions are disabled", async () => {
     const result = await handleCodexAppServerElicitationRequest({
       requestParams: buildConnectorPluginApprovalElicitation(),

extensions/codex/src/app-server/elicitation-bridge.ts

@@ -9,7 +9,6 @@ import {
   mapExecDecisionToOutcome,
   requestPluginApproval,
   type AppServerApprovalOutcome,
-  type ExecApprovalDecision,
   waitForPluginApprovalDecision,
 } from "./plugin-approval-roundtrip.js";
 import type {
@@ -29,8 +28,6 @@ type BridgeableApprovalElicitation = {
   description: string;
   requestedSchema: JsonObject;
   meta: JsonObject;
-  persistHintsMode?: "legacy" | "explicit";
-  allowedDecisions?: ExecApprovalDecision[];
 };
 
 type PluginElicitationResolution =
@@ -114,12 +111,7 @@ export async function handleCodexAppServerElicitationRequest(params: {
       logPluginElicitationDecline("missing_active_turn", requestParams);
       return declineElicitationResponse();
     }
-    return await buildPluginPolicyElicitationResponse({
-      entry: pluginResolution.entry,
-      requestParams,
-      paramsForRun: params.paramsForRun,
-      signal: params.signal,
-    });
+    return buildPluginPolicyElicitationResponse(pluginResolution.entry, requestParams);
   }
 
   const approvalPrompt =
@@ -133,10 +125,9 @@ export async function handleCodexAppServerElicitationRequest(params: {
     paramsForRun: params.paramsForRun,
     title: approvalPrompt.title,
     description: approvalPrompt.description,
-    allowedDecisions: approvalPrompt.allowedDecisions,
     signal: params.signal,
   });
-  return buildElicitationResponse(approvalPrompt, outcome);
+  return buildElicitationResponse(approvalPrompt.requestedSchema, approvalPrompt.meta, outcome);
 }
 
 function matchesCurrentThread(requestParams: JsonObject | undefined, threadId: string): boolean {
@@ -293,104 +284,28 @@ function normalizePluginIdentityText(value: string): string {
   return value.toLowerCase().replace(/[^a-z0-9]+/g, "");
 }
 
-async function buildPluginPolicyElicitationResponse(params: {
-  entry: PluginAppPolicyContextEntry;
-  requestParams: JsonObject;
-  paramsForRun: EmbeddedRunAttemptParams;
-  signal?: AbortSignal;
-}): Promise<JsonValue> {
-  const mode = resolvePluginDestructiveApprovalMode(params.entry);
-  if (mode === "deny") {
-    logPluginElicitationDecline("destructive_actions_disabled", params.requestParams);
-    return declineElicitationResponse();
-  }
-  const approvalPrompt = readPluginApprovalElicitation(params.entry, params.requestParams);
-  if (!approvalPrompt) {
-    logPluginElicitationDecline("unsupported_schema", params.requestParams);
-    return declineElicitationResponse();
-  }
-  const response = buildElicitationResponse(approvalPrompt, "approved-once");
-  if (isJsonObject(response) && response.action === "accept") {
-    if (mode === "auto") {
-      return response;
-    }
-    const outcome = await requestPluginApprovalOutcome({
-      paramsForRun: params.paramsForRun,
-      title: approvalPrompt.title,
-      description: approvalPrompt.description,
-      allowedDecisions: approvalPrompt.allowedDecisions,
-      signal: params.signal,
-    });
-    return buildElicitationResponse(approvalPrompt, outcome);
-  }
-  logPluginElicitationDecline("unmappable_schema", params.requestParams);
-  return declineElicitationResponse();
-}
-
-function resolvePluginDestructiveApprovalMode(
-  entry: PluginAppPolicyContextEntry,
-): "auto" | "deny" | "on-request" {
-  return entry.destructiveApprovalMode ?? (entry.allowDestructiveActions ? "auto" : "deny");
-}
-
-function readPluginApprovalElicitation(
+function buildPluginPolicyElicitationResponse(
   entry: PluginAppPolicyContextEntry,
   requestParams: JsonObject,
-): BridgeableApprovalElicitation | undefined {
+): JsonValue {
+  if (!entry.allowDestructiveActions) {
+    logPluginElicitationDecline("destructive_actions_disabled", requestParams);
+    return declineElicitationResponse();
+  }
   if (
     readString(requestParams, "mode") !== "form" ||
     !isJsonObject(requestParams.requestedSchema)
   ) {
-    return undefined;
+    logPluginElicitationDecline("unsupported_schema", requestParams);
+    return declineElicitationResponse();
   }
-  const requestedSchema = requestParams.requestedSchema;
   const meta = isJsonObject(requestParams["_meta"]) ? requestParams["_meta"] : {};
-  const title =
-    sanitizeDisplayText(readString(requestParams, "message") ?? "") || "Codex plugin approval";
-  const descriptionMeta: JsonObject = { ...meta };
-  if (!readString(descriptionMeta, MCP_TOOL_APPROVAL_CONNECTOR_NAME_KEY)) {
-    descriptionMeta[MCP_TOOL_APPROVAL_CONNECTOR_NAME_KEY] = entry.pluginName;
+  const response = buildElicitationResponse(requestParams.requestedSchema, meta, "approved-once");
+  if (isJsonObject(response) && response.action === "accept") {
+    return response;
   }
-  return {
-    title,
-    description: buildApprovalDescription({
-      title,
-      meta: descriptionMeta,
-      requestedSchema,
-      serverName: sanitizeOptionalDisplayText(readString(requestParams, "serverName")),
-    }),
-    requestedSchema,
-    meta,
-    persistHintsMode: "explicit",
-    allowedDecisions: buildApprovalAllowedDecisions(requestedSchema, meta),
-  };
-}
-
-function buildApprovalAllowedDecisions(
-  requestedSchema: JsonObject,
-  meta: JsonObject,
-): ExecApprovalDecision[] {
-  return canMapPersistentApproval(requestedSchema, meta)
-    ? ["allow-once", "allow-always", "deny"]
-    : ["allow-once", "deny"];
-}
-
-function canMapPersistentApproval(requestedSchema: JsonObject, meta: JsonObject): boolean {
-  const persistHints = readPersistHints(meta, "explicit");
-  if (persistHints.length > 0) {
-    return persistHints.includes("always");
-  }
-  const properties = isJsonObject(requestedSchema.properties) ? requestedSchema.properties : {};
-  return Object.entries(properties).some(([name, value]) => {
-    const schema = isJsonObject(value) ? value : undefined;
-    if (!schema) {
-      return false;
-    }
-    return (
-      isPersistField({ name, schema, required: false }) &&
-      chooseAlwaysPersistOptionValue(readEnumOptions(schema)) !== undefined
-    );
-  });
+  logPluginElicitationDecline("unmappable_schema", requestParams);
+  return declineElicitationResponse();
 }
 
 function declineElicitationResponse(): JsonValue {
@@ -643,7 +558,6 @@ async function requestPluginApprovalOutcome(params: {
   paramsForRun: EmbeddedRunAttemptParams;
   title: string;
   description: string;
-  allowedDecisions?: ExecApprovalDecision[];
   signal?: AbortSignal;
 }): Promise<AppServerApprovalOutcome> {
   try {
@@ -653,7 +567,6 @@ async function requestPluginApprovalOutcome(params: {
       description: params.description,
       severity: "warning",
       toolName: "codex_mcp_tool_approval",
-      allowedDecisions: params.allowedDecisions,
     });
 
     const approvalId = requestResult?.id;
@@ -671,13 +584,10 @@ async function requestPluginApprovalOutcome(params: {
 }
 
 function buildElicitationResponse(
-  approvalPrompt: Pick<
-    BridgeableApprovalElicitation,
-    "requestedSchema" | "meta" | "persistHintsMode"
-  >,
+  requestedSchema: JsonObject,
+  meta: JsonObject,
   outcome: AppServerApprovalOutcome,
 ): JsonValue {
-  const { requestedSchema, meta } = approvalPrompt;
   if (outcome === "cancelled") {
     return { action: "cancel", content: null, _meta: null };
   }
@@ -685,13 +595,13 @@ function buildElicitationResponse(
     return { action: "decline", content: null, _meta: null };
   }
 
-  const content = buildAcceptedContent(approvalPrompt, outcome);
+  const content = buildAcceptedContent(requestedSchema, meta, outcome);
   if (!content) {
     if (hasNoSchemaProperties(requestedSchema)) {
       return {
         action: "accept",
         content: null,
-        _meta: buildAcceptedMeta(meta, outcome, approvalPrompt.persistHintsMode ?? "legacy"),
+        _meta: buildAcceptedMeta(meta, outcome),
       };
     }
     embeddedAgentLog.warn("codex MCP approval elicitation approved without a mappable response", {
@@ -701,21 +611,14 @@ function buildElicitationResponse(
     });
     return { action: "decline", content: null, _meta: null };
   }
-  return {
-    action: "accept",
-    content,
-    _meta: buildAcceptedMeta(meta, outcome, approvalPrompt.persistHintsMode ?? "legacy"),
-  };
+  return { action: "accept", content, _meta: buildAcceptedMeta(meta, outcome) };
 }
 
 function buildAcceptedContent(
-  approvalPrompt: Pick<
-    BridgeableApprovalElicitation,
-    "requestedSchema" | "meta" | "persistHintsMode"
-  >,
+  requestedSchema: JsonObject,
+  meta: JsonObject,
   outcome: AppServerApprovalOutcome,
 ): JsonObject | undefined {
-  const { requestedSchema, meta } = approvalPrompt;
   const properties = isJsonObject(requestedSchema.properties)
     ? requestedSchema.properties
     : undefined;
@@ -738,7 +641,7 @@ function buildAcceptedContent(
     const property = { name, schema, required: required.has(name) };
     const next =
       readApprovalFieldValue(property, outcome) ??
-      readPersistFieldValue(property, meta, outcome, approvalPrompt.persistHintsMode ?? "legacy") ??
+      readPersistFieldValue(property, meta, outcome) ??
       readFallbackFieldValue(property, outcome);
 
     if (next === undefined) {
@@ -788,12 +691,11 @@ function readPersistFieldValue(
   property: ApprovalPropertyContext,
   meta: JsonObject,
   outcome: AppServerApprovalOutcome,
-  persistHintsMode: "legacy" | "explicit",
 ): JsonValue | undefined {
   if (!isPersistField(property) || outcome !== "approved-session") {
     return undefined;
   }
-  const persistHints = readPersistHints(meta, persistHintsMode);
+  const persistHints = readPersistHints(meta);
   const options = readEnumOptions(property.schema);
   if (options.length === 0) {
     return undefined;
@@ -805,9 +707,6 @@ function readPersistFieldValue(
     );
     return match?.value;
   }
-  if (persistHintsMode === "explicit") {
-    return chooseAlwaysPersistOptionValue(options);
-  }
   return undefined;
 }
 
@@ -845,7 +744,7 @@ function propertyText(property: ApprovalPropertyContext): string {
     .join(" ");
 }
 
-function readPersistHints(meta: JsonObject, mode: "legacy" | "explicit" = "legacy"): string[] {
+function readPersistHints(meta: JsonObject): string[] {
   const raw = meta.persist;
   if (typeof raw === "string") {
     return [raw];
@@ -853,18 +752,14 @@ function readPersistHints(meta: JsonObject, mode: "legacy" | "explicit" = "legac
   if (Array.isArray(raw)) {
     return raw.filter((entry): entry is string => typeof entry === "string");
   }
-  return mode === "legacy" ? ["session", "always"] : [];
+  return ["session", "always"];
 }
 
-function buildAcceptedMeta(
-  meta: JsonObject,
-  outcome: AppServerApprovalOutcome,
-  persistHintsMode: "legacy" | "explicit",
-): JsonObject | null {
+function buildAcceptedMeta(meta: JsonObject, outcome: AppServerApprovalOutcome): JsonObject | null {
   if (outcome !== "approved-session") {
     return null;
   }
-  const persist = choosePersistHint(readPersistHints(meta, persistHintsMode));
+  const persist = choosePersistHint(readPersistHints(meta));
   return persist ? { persist } : null;
 }
 
@@ -878,20 +773,6 @@ function choosePersistHint(persistHints: string[]): "always" | "session" | undef
   return undefined;
 }
 
-function chooseAlwaysPersistOptionValue(
-  options: Array<{ value: string; label: string }>,
-): string | undefined {
-  const always = options.find((option) => optionMatchesPersist(option, "always"));
-  return always?.value;
-}
-
-function optionMatchesPersist(
-  option: { value: string; label: string },
-  persist: "always" | "session",
-): boolean {
-  return option.value.toLowerCase() === persist || option.label.toLowerCase() === persist;
-}
-
 function hasNoSchemaProperties(requestedSchema: JsonObject): boolean {
   const properties = isJsonObject(requestedSchema.properties) ? requestedSchema.properties : {};
   return Object.keys(properties).length === 0;

extensions/codex/src/app-server/plugin-activation.test.ts

@@ -303,7 +303,6 @@ function identity(pluginName: string): ResolvedCodexPluginPolicy {
     pluginName,
     enabled: true,
     allowDestructiveActions: false,
-    destructiveApprovalMode: "deny",
   };
 }
 

extensions/codex/src/app-server/plugin-approval-roundtrip.ts

@@ -12,7 +12,7 @@ const DEFAULT_CODEX_APPROVAL_TIMEOUT_MS = 120_000;
 const MAX_PLUGIN_APPROVAL_TITLE_LENGTH = 80;
 const MAX_PLUGIN_APPROVAL_DESCRIPTION_LENGTH = 256;
 
-export type ExecApprovalDecision = "allow-once" | "allow-always" | "deny";
+type ExecApprovalDecision = "allow-once" | "allow-always" | "deny";
 
 /** Normalized Codex app-server approval outcome after a gateway decision. */
 export type AppServerApprovalOutcome =
@@ -40,7 +40,6 @@ export async function requestPluginApproval(params: {
   severity: "info" | "warning";
   toolName: string;
   toolCallId?: string;
-  allowedDecisions?: ExecApprovalDecision[];
 }): Promise<ApprovalRequestResult | undefined> {
   const timeoutMs = DEFAULT_CODEX_APPROVAL_TIMEOUT_MS;
   return callGatewayTool(
@@ -61,7 +60,6 @@ export async function requestPluginApproval(params: {
       turnSourceThreadId: params.paramsForRun.currentThreadTs,
       timeoutMs,
       twoPhase: true,
-      ...(params.allowedDecisions ? { allowedDecisions: params.allowedDecisions } : {}),
     },
     { expectFinal: false },
   ) as Promise<ApprovalRequestResult | undefined>;

extensions/codex/src/app-server/plugin-thread-config.test.ts

@@ -73,7 +73,6 @@ describe("Codex plugin thread config", () => {
       marketplaceName: CODEX_PLUGINS_MARKETPLACE_NAME,
       pluginName: "google-calendar",
       allowDestructiveActions: true,
-      destructiveApprovalMode: "auto",
       mcpServerNames: ["google-calendar"],
     });
     expect(config.diagnostics).toStrictEqual([]);
@@ -108,9 +107,6 @@ describe("Codex plugin thread config", () => {
     expect(
       pluginOverrideDisabled.policyContext.apps["google-calendar-app"]?.allowDestructiveActions,
     ).toBe(false);
-    expect(
-      pluginOverrideDisabled.policyContext.apps["google-calendar-app"]?.destructiveApprovalMode,
-    ).toBe("deny");
 
     const pluginOverrideEnabled = await buildReadyGoogleCalendarThreadConfig({
       codexPlugins: {
@@ -138,36 +134,6 @@ describe("Codex plugin thread config", () => {
     expect(
       pluginOverrideEnabled.policyContext.apps["google-calendar-app"]?.allowDestructiveActions,
     ).toBe(true);
-    expect(
-      pluginOverrideEnabled.policyContext.apps["google-calendar-app"]?.destructiveApprovalMode,
-    ).toBe("auto");
-  });
-
-  it("exposes destructive app access while marking on-request approval mode", async () => {
-    const config = await buildReadyGoogleCalendarThreadConfig({
-      codexPlugins: {
-        enabled: true,
-        allow_destructive_actions: "on-request",
-        plugins: {
-          "google-calendar": {
-            marketplaceName: CODEX_PLUGINS_MARKETPLACE_NAME,
-            pluginName: "google-calendar",
-          },
-        },
-      },
-    });
-
-    const apps = config.configPatch?.apps as Record<string, unknown> | undefined;
-    expect(apps?.["google-calendar-app"]).toEqual({
-      enabled: true,
-      destructive_enabled: true,
-      open_world_enabled: true,
-      default_tools_approval_mode: "auto",
-    });
-    expect(config.policyContext.apps["google-calendar-app"]).toMatchObject({
-      allowDestructiveActions: true,
-      destructiveApprovalMode: "on-request",
-    });
   });
 
   it("builds a restrictive app config when native plugin support is disabled", async () => {
@@ -301,7 +267,6 @@ describe("Codex plugin thread config", () => {
       marketplaceName: CODEX_PLUGINS_MARKETPLACE_NAME,
       pluginName: "google-calendar",
       allowDestructiveActions: true,
-      destructiveApprovalMode: "auto",
       mcpServerNames: [],
     });
     expect(config.diagnostics).toStrictEqual([]);
@@ -373,7 +338,6 @@ describe("Codex plugin thread config", () => {
           pluginName: "google-calendar",
           enabled: true,
           allowDestructiveActions: true,
-          destructiveApprovalMode: "auto",
         },
         message: "google-calendar-app is not accessible or enabled for google-calendar.",
       },
@@ -444,7 +408,6 @@ describe("Codex plugin thread config", () => {
       marketplaceName: CODEX_PLUGINS_MARKETPLACE_NAME,
       pluginName: "google-calendar",
       allowDestructiveActions: true,
-      destructiveApprovalMode: "auto",
       mcpServerNames: [],
     });
     expect(config.diagnostics).toStrictEqual([]);
@@ -535,7 +498,6 @@ describe("Codex plugin thread config", () => {
       marketplaceName: CODEX_PLUGINS_MARKETPLACE_NAME,
       pluginName: "google-calendar",
       allowDestructiveActions: true,
-      destructiveApprovalMode: "auto",
       mcpServerNames: [],
     });
     expect(config.diagnostics).toStrictEqual([]);

extensions/codex/src/app-server/plugin-thread-config.ts

@@ -13,7 +13,6 @@ import {
 } from "./app-inventory-cache.js";
 import {
   resolveCodexPluginsPolicy,
-  type CodexPluginDestructiveApprovalMode,
   type ResolvedCodexPluginPolicy,
   type ResolvedCodexPluginsPolicy,
 } from "./config.js";
@@ -37,7 +36,6 @@ export type PluginAppPolicyContextEntry = {
   marketplaceName: ResolvedCodexPluginPolicy["marketplaceName"];
   pluginName: string;
   allowDestructiveActions: boolean;
-  destructiveApprovalMode?: CodexPluginDestructiveApprovalMode;
   mcpServerNames: string[];
 };
 
@@ -248,7 +246,6 @@ export async function buildCodexPluginThreadConfig(
         marketplaceName: record.policy.marketplaceName,
         pluginName: record.policy.pluginName,
         allowDestructiveActions: record.policy.allowDestructiveActions,
-        destructiveApprovalMode: record.policy.destructiveApprovalMode,
         mcpServerNames: [...(record.detail?.mcpServers ?? [])].toSorted(),
       };
     }
@@ -428,14 +425,12 @@ function policyFingerprint(policy: ResolvedCodexPluginsPolicy): JsonValue {
   return {
     enabled: policy.enabled,
     allowDestructiveActions: policy.allowDestructiveActions,
-    destructiveApprovalMode: policy.destructiveApprovalMode,
     plugins: policy.pluginPolicies.map((plugin) => ({
       configKey: plugin.configKey,
       marketplaceName: plugin.marketplaceName,
       pluginName: plugin.pluginName,
       enabled: plugin.enabled,
       allowDestructiveActions: plugin.allowDestructiveActions,
-      destructiveApprovalMode: plugin.destructiveApprovalMode,
     })),
   };
 }

extensions/codex/src/app-server/run-attempt-test-harness.ts

@@ -10,6 +10,7 @@ import {
 } from "openclaw/plugin-sdk/agent-harness-runtime";
 import { resetDiagnosticEventsForTest } from "openclaw/plugin-sdk/diagnostic-runtime";
 import { clearInternalHooks, resetGlobalHookRunner } from "openclaw/plugin-sdk/hook-runtime";
+import { clearMemoryPluginState } from "openclaw/plugin-sdk/memory-core-host-runtime-core";
 import { clearPluginCommands } from "openclaw/plugin-sdk/plugin-runtime";
 import { resolvePreferredOpenClawTmpDir } from "openclaw/plugin-sdk/temp-path";
 import { afterEach, beforeEach, expect, vi } from "vitest";
@@ -495,6 +496,7 @@ export function setupRunAttemptTestHooks(): void {
   beforeEach(async () => {
     vi.useRealTimers();
     clearInternalHooks();
+    clearMemoryPluginState();
     resetAgentEventsForTest();
     resetDiagnosticEventsForTest();
     vi.stubEnv("OPENCLAW_TRAJECTORY", "0");
@@ -512,6 +514,7 @@ export function setupRunAttemptTestHooks(): void {
     testing.clearPendingCodexNativeHookRelayUnregistersForTests();
     resetCodexRateLimitCacheForTests();
     nativeHookRelayTesting.clearNativeHookRelaysForTests();
+    clearMemoryPluginState();
     clearPluginCommands();
     resetAgentEventsForTest();
     resetDiagnosticEventsForTest();

extensions/codex/src/app-server/run-attempt.test.ts

@@ -12,6 +12,7 @@ import {
   type DiagnosticEventPayload,
 } from "openclaw/plugin-sdk/diagnostic-runtime";
 import { initializeGlobalHookRunner, registerInternalHook } from "openclaw/plugin-sdk/hook-runtime";
+import { registerMemoryCapability } from "openclaw/plugin-sdk/memory-core-host-runtime-core";
 import { registerPluginCommand } from "openclaw/plugin-sdk/plugin-runtime";
 import { createMockPluginRegistry } from "openclaw/plugin-sdk/plugin-test-runtime";
 import { describe, expect, it, vi } from "vitest";
@@ -397,6 +398,37 @@ function createRuntimeDynamicTool(name: string): RuntimeDynamicToolForTest {
   };
 }
 
+function registerMemoryPromptForTest() {
+  registerMemoryCapability("memory-core", {
+    promptBuilder({ availableTools }) {
+      const hasMemorySearch = availableTools.has("memory_search");
+      const hasMemoryGet = availableTools.has("memory_get");
+      if (hasMemorySearch && hasMemoryGet) {
+        return [
+          "## Memory Recall",
+          "Test recall: run memory_search on MEMORY.md + memory/*.md + indexed session transcripts; then use memory_get.",
+          "",
+        ];
+      }
+      if (hasMemorySearch) {
+        return [
+          "## Memory Recall",
+          "Test recall: run memory_search on MEMORY.md + memory/*.md + indexed session transcripts.",
+          "",
+        ];
+      }
+      if (hasMemoryGet) {
+        return [
+          "## Memory Recall",
+          "Test recall: run memory_get for a specific memory file or note.",
+          "",
+        ];
+      }
+      return [];
+    },
+  });
+}
+
 function buildEmptyCodexToolTelemetry(): CodexAppServerToolTelemetry {
   return {
     didSendViaMessagingTool: false,
@@ -2203,6 +2235,7 @@ describe("runCodexAppServerAttempt", () => {
     await fs.writeFile(path.join(workspaceDir, "TOOLS.md"), toolGuidance);
     await fs.writeFile(path.join(workspaceDir, "USER.md"), userProfile);
     await fs.writeFile(path.join(workspaceDir, "MEMORY.md"), memorySummary);
+    registerMemoryPromptForTest();
     testing.setOpenClawCodingToolsFactoryForTests(() => [
       createRuntimeDynamicTool("memory_search"),
       createRuntimeDynamicTool("memory_get"),
@@ -2236,12 +2269,20 @@ describe("runCodexAppServerAttempt", () => {
     expect(collaborationInstructions).toContain(identityGuidance);
     expect(collaborationInstructions).not.toContain(toolGuidance);
     expect(collaborationInstructions).toContain(userProfile);
+    expect(collaborationInstructions).toContain("## Memory Recall");
+    expect(collaborationInstructions).toContain("MEMORY.md + memory/*.md");
     expect(collaborationInstructions).toContain("OpenClaw Workspace Memory");
     expect(collaborationInstructions).toContain(
       "MEMORY.md exists in the active agent workspace as a memory file, not an instruction file",
     );
     expect(collaborationInstructions).toContain("memory_search");
     expect(collaborationInstructions).toContain("memory_get");
+    expect(collaborationInstructions).toContain(
+      "When the memory guidance above calls for memory recall, use an already-loaded memory tool directly.",
+    );
+    expect(collaborationInstructions).toContain(
+      "If the needed memory tool is deferred and not currently callable, use `tool_search` to load it, then call that memory tool.",
+    );
     expect(collaborationInstructions).not.toContain(memorySummary);
     expect(inputText).not.toContain("OpenClaw runtime context for this turn:");
     expect(inputText).not.toContain("does not override Codex system/developer instructions");
@@ -2297,6 +2338,65 @@ describe("runCodexAppServerAttempt", () => {
     });
   });
 
+  it("adds memory recall guidance when dated memory notes exist without root MEMORY.md", async () => {
+    const sessionFile = path.join(tempDir, "session.jsonl");
+    const workspaceDir = path.join(tempDir, "workspace");
+    const datedMemory = "User avoids Chase cards while over 5/24.";
+    await fs.mkdir(path.join(workspaceDir, "memory"), { recursive: true });
+    await fs.writeFile(path.join(workspaceDir, "memory/2026-06-09.md"), datedMemory);
+    registerMemoryPromptForTest();
+    testing.setOpenClawCodingToolsFactoryForTests(() => [
+      createRuntimeDynamicTool("memory_search"),
+      createRuntimeDynamicTool("memory_get"),
+    ]);
+    const params = createParams(sessionFile, workspaceDir);
+    params.disableTools = false;
+    params.runtimePlan = createCodexRuntimePlanFixture();
+    setAgentWorkspaceForTest(params, workspaceDir);
+
+    const { collaborationInstructions, inputText } = await buildCodexTurnContextForTest(
+      params,
+      workspaceDir,
+    );
+
+    expect(collaborationInstructions).toContain("## Memory Recall");
+    expect(collaborationInstructions).toContain("MEMORY.md + memory/*.md");
+    expect(collaborationInstructions).toContain("memory_search");
+    expect(collaborationInstructions).toContain("memory_get");
+    expect(collaborationInstructions).not.toContain("OpenClaw Workspace Memory");
+    expect(collaborationInstructions).not.toContain(datedMemory);
+    expect(inputText).toBe("hello");
+    expect(inputText).not.toContain(datedMemory);
+  });
+
+  it("does not synthesize memory recall guidance without a registered memory prompt builder", async () => {
+    const sessionFile = path.join(tempDir, "session.jsonl");
+    const workspaceDir = path.join(tempDir, "workspace");
+    const memorySummary = "User avoids Chase cards while over 5/24.";
+    await fs.mkdir(workspaceDir, { recursive: true });
+    await fs.writeFile(path.join(workspaceDir, "MEMORY.md"), memorySummary);
+    testing.setOpenClawCodingToolsFactoryForTests(() => [
+      createRuntimeDynamicTool("memory_search"),
+      createRuntimeDynamicTool("memory_get"),
+    ]);
+    const params = createParams(sessionFile, workspaceDir);
+    params.disableTools = false;
+    params.runtimePlan = createCodexRuntimePlanFixture();
+    setAgentWorkspaceForTest(params, workspaceDir);
+
+    const { collaborationInstructions, inputText } = await buildCodexTurnContextForTest(
+      params,
+      workspaceDir,
+    );
+
+    expect(collaborationInstructions).not.toContain("## Memory Recall");
+    expect(collaborationInstructions).toContain("OpenClaw Workspace Memory");
+    expect(collaborationInstructions).not.toContain("Use `tool_search` first");
+    expect(collaborationInstructions).not.toContain(memorySummary);
+    expect(inputText).toBe("hello");
+    expect(inputText).not.toContain(memorySummary);
+  });
+
   it("sends workspace bootstrap instructions through Codex app-server payloads", async () => {
     const sessionFile = path.join(tempDir, "session.jsonl");
     const workspaceDir = path.join(tempDir, "workspace");
@@ -2405,6 +2505,7 @@ describe("runCodexAppServerAttempt", () => {
     const memorySummary = "Memory summary goes here.";
     await fs.mkdir(workspaceDir, { recursive: true });
     await fs.writeFile(path.join(workspaceDir, "MEMORY.md"), memorySummary);
+    registerMemoryPromptForTest();
     testing.setOpenClawCodingToolsFactoryForTests(() => [createRuntimeDynamicTool("memory_get")]);
     const params = createParams(sessionFile, workspaceDir);
     params.disableTools = false;
@@ -2417,6 +2518,7 @@ describe("runCodexAppServerAttempt", () => {
     expect(inputText).not.toContain("memory_get");
     expect(inputText).not.toContain("memory_search");
     expect(inputText).not.toContain(memorySummary);
+    expect(collaborationInstructions).toContain("## Memory Recall");
     expect(collaborationInstructions).toContain("OpenClaw Workspace Memory");
     expect(collaborationInstructions).toContain("memory_get");
     expect(collaborationInstructions).not.toContain("memory_search");
@@ -2595,6 +2697,7 @@ describe("runCodexAppServerAttempt", () => {
     const memorySummary = "Memory summary goes here.";
     await fs.mkdir(workspaceDir, { recursive: true });
     await fs.writeFile(path.join(workspaceDir, "MEMORY.md"), memorySummary);
+    registerMemoryPromptForTest();
     testing.setOpenClawCodingToolsFactoryForTests(() => [
       createRuntimeDynamicTool("memory_search"),
       createRuntimeDynamicTool("memory_get"),
@@ -2604,10 +2707,10 @@ describe("runCodexAppServerAttempt", () => {
     params.runtimePlan = createCodexRuntimePlanFixture();
     setAgentWorkspaceForTest(params, path.join(tempDir, "memory-workspace"));
 
-    const { inputText, systemPromptReport } = await buildCodexTurnContextForTest(
-      params,
-      workspaceDir,
-    );
+    const { collaborationInstructions, inputText, systemPromptReport } =
+      await buildCodexTurnContextForTest(params, workspaceDir);
+    expect(collaborationInstructions).not.toContain("## Memory Recall");
+    expect(collaborationInstructions).not.toContain("OpenClaw Workspace Memory");
     expect(inputText).not.toContain("OpenClaw Workspace Memory");
     expect(inputText).toContain(memorySummary);
 

extensions/codex/src/app-server/session-binding.test.ts

@@ -108,35 +108,6 @@ describe("codex app-server session binding", () => {
     expect(binding?.pluginAppPolicyContext).toEqual(pluginAppPolicyContext);
   });
 
-  it("round-trips plugin app policy context destructive approval mode", async () => {
-    const sessionFile = path.join(tempDir, "session.json");
-    const pluginAppPolicyContext = {
-      fingerprint: "plugin-policy-1",
-      apps: {
-        "google-calendar-app": {
-          configKey: "google-calendar",
-          marketplaceName: "openai-curated" as const,
-          pluginName: "google-calendar",
-          allowDestructiveActions: true,
-          destructiveApprovalMode: "on-request" as const,
-          mcpServerNames: ["google-calendar"],
-        },
-      },
-      pluginAppIds: {
-        "google-calendar": ["google-calendar-app"],
-      },
-    };
-    await writeCodexAppServerBinding(sessionFile, {
-      threadId: "thread-123",
-      cwd: tempDir,
-      pluginAppPolicyContext,
-    });
-
-    const binding = await readCodexAppServerBinding(sessionFile);
-
-    expect(binding?.pluginAppPolicyContext).toEqual(pluginAppPolicyContext);
-  });
-
   it("round-trips context-engine binding metadata", async () => {
     const sessionFile = path.join(tempDir, "session.json");
     await writeCodexAppServerBinding(sessionFile, {

extensions/codex/src/app-server/session-binding.ts

@@ -333,7 +333,6 @@ function readPluginAppPolicyContext(value: unknown): PluginAppPolicyContext | un
       entry.marketplaceName !== CODEX_PLUGINS_MARKETPLACE_NAME ||
       typeof entry.pluginName !== "string" ||
       typeof entry.allowDestructiveActions !== "boolean" ||
-      !isValidDestructiveApprovalMode(entry.destructiveApprovalMode) ||
       !Array.isArray(entry.mcpServerNames) ||
       entry.mcpServerNames.some((serverName) => typeof serverName !== "string")
     ) {
@@ -344,9 +343,6 @@ function readPluginAppPolicyContext(value: unknown): PluginAppPolicyContext | un
       marketplaceName: entry.marketplaceName,
       pluginName: entry.pluginName,
       allowDestructiveActions: entry.allowDestructiveActions,
-      ...(entry.destructiveApprovalMode
-        ? { destructiveApprovalMode: entry.destructiveApprovalMode }
-        : {}),
       mcpServerNames: entry.mcpServerNames,
     };
   }
@@ -370,12 +366,6 @@ function readPluginAppPolicyContext(value: unknown): PluginAppPolicyContext | un
   };
 }
 
-function isValidDestructiveApprovalMode(
-  value: unknown,
-): value is PluginAppPolicyContext["apps"][string]["destructiveApprovalMode"] | undefined {
-  return value === undefined || value === "auto" || value === "deny" || value === "on-request";
-}
-
 /** Removes the Codex app-server binding sidecar if present. */
 export async function clearCodexAppServerBinding(
   sessionFile: string,

extensions/codex/src/command-plugins-management.ts

@@ -23,7 +23,7 @@ export type CodexPluginConfigEntry = {
   enabled?: boolean;
   marketplaceName?: string;
   pluginName?: string;
-  allow_destructive_actions?: boolean | "on-request";
+  allow_destructive_actions?: boolean;
 };
 
 export type CodexPluginsConfigBlock = {

extensions/codex/src/migration/apply.ts

@@ -508,7 +508,6 @@ function readCodexPluginPolicy(item: MigrationItem): ResolvedCodexPluginPolicy |
     pluginName,
     enabled: true,
     allowDestructiveActions: true,
-    destructiveApprovalMode: "auto",
   };
 }
 

extensions/codex/src/migration/plan.ts

@@ -258,12 +258,12 @@ export function readCodexPluginMigrationConfigEntry(
 
 function readExistingAllowDestructiveActions(
   config: MigrationProviderContext["config"],
-): boolean | "on-request" | undefined {
+): boolean | undefined {
   const value = readMigrationConfigPath(config as Record<string, unknown>, [
     ...CODEX_PLUGIN_NATIVE_CONFIG_PATH,
     "allow_destructive_actions",
   ]);
-  return value === "on-request" ? "on-request" : asBoolean(value);
+  return asBoolean(value);
 }
 
 export function buildCodexPluginsConfigValue(

extensions/diagnostics-otel/src/service.test.ts

@@ -171,6 +171,7 @@ import {
   type DiagnosticEventPrivateData,
 } from "openclaw/plugin-sdk/diagnostic-runtime";
 import {
+  emitDiagnosticEventWithTrustedTraceContext,
   emitInternalDiagnosticEventForTest,
   logMessageDispatchStarted,
   logMessageProcessed,
@@ -362,7 +363,11 @@ function histogramCreateOptions(name: string) {
 
 async function emitAndCaptureLog(
   event: Omit<Extract<Parameters<typeof emitDiagnosticEvent>[0], { type: "log.record" }>, "type">,
-  options: { captureContent?: OtelContextFlags["captureContent"]; trusted?: boolean } = {},
+  options: {
+    captureContent?: OtelContextFlags["captureContent"];
+    trusted?: boolean;
+    trustedTraceContext?: boolean;
+  } = {},
 ) {
   const service = createDiagnosticsOtelService();
   const ctx = createOtelContext(OTEL_TEST_ENDPOINT, {
@@ -370,7 +375,11 @@ async function emitAndCaptureLog(
     ...(options.captureContent !== undefined ? { captureContent: options.captureContent } : {}),
   });
   await service.start(ctx);
-  const emit = options.trusted ? emitTrustedDiagnosticEvent : emitDiagnosticEvent;
+  const emit = options.trusted
+    ? emitTrustedDiagnosticEvent
+    : options.trustedTraceContext
+      ? emitDiagnosticEventWithTrustedTraceContext
+      : emitDiagnosticEvent;
   emit({
     type: "log.record",
     ...event,
@@ -1391,6 +1400,28 @@ describe("diagnostics-otel service", () => {
     expect(emitCall?.context).toBeUndefined();
   });
 
+  test("attaches trace-only trusted context to exported logs", async () => {
+    const emitCall = await emitAndCaptureLog(
+      {
+        level: "INFO",
+        message: "traceable log",
+        trace: {
+          traceId: TRACE_ID,
+          spanId: SPAN_ID,
+          traceFlags: "01",
+        },
+      },
+      { trustedTraceContext: true },
+    );
+
+    expect(emitCall?.body).toBe("log");
+    expect(telemetryState.tracer.setSpanContext).toHaveBeenCalledTimes(1);
+    const emitContext = emitCall?.context as { spanContext?: Record<string, unknown> } | undefined;
+    const emitSpanContext = emitContext?.spanContext;
+    expect(emitSpanContext?.traceId).toBe(TRACE_ID);
+    expect(emitSpanContext?.spanId).toBe(SPAN_ID);
+  });
+
   test("attaches trusted diagnostic trace context to exported logs", async () => {
     const emitCall = await emitAndCaptureLog(
       {

extensions/diagnostics-otel/src/service.ts

@@ -1031,7 +1031,9 @@ function contextForTrustedTraceContext(
   evt: DiagnosticEventPayload,
   metadata: DiagnosticEventMetadata,
 ) {
-  return metadata.trusted ? contextForTraceContext(evt.trace) : undefined;
+  return metadata.trusted || metadata.trustedTraceContext === true
+    ? contextForTraceContext(evt.trace)
+    : undefined;
 }
 
 function addTraceAttributes(
@@ -1626,7 +1628,7 @@ export function createDiagnosticsOtelService(): OpenClawPluginService {
             if (evt.code?.functionName) {
               assignOtelLogAttribute(attributes, "code.function", evt.code.functionName);
             }
-            if (metadata.trusted) {
+            if (metadata.trusted || metadata.trustedTraceContext === true) {
               addTraceAttributes(attributes, evt.trace);
             }
 

extensions/moonshot/index.test.ts

@@ -89,4 +89,60 @@ describe("moonshot provider plugin", () => {
       thinking: { type: "disabled" },
     });
   });
+
+  it("keeps Kimi K2.7 Code thinking always on without sending a thinking field", async () => {
+    const provider = await registerSingleProviderPlugin(plugin);
+    const capturedStream = createCapturedThinkingConfigStream();
+
+    const wrapped = provider.wrapSimpleCompletionStreamFn?.({
+      provider: "moonshot",
+      modelId: "kimi-k2.7-code",
+      thinkingLevel: "off",
+      streamFn: capturedStream.streamFn,
+    } as never);
+
+    void wrapped?.(
+      {
+        api: "openai-completions",
+        provider: "moonshot",
+        id: "kimi-k2.7-code",
+      } as Model<"openai-completions">,
+      { messages: [] } as Context,
+      {},
+    );
+
+    expect(capturedStream.getCapturedPayload()).toEqual({
+      config: { thinkingConfig: { thinkingBudget: -1 } },
+    });
+    expect(
+      provider.wrapSimpleCompletionStreamFn?.({
+        provider: "moonshot",
+        modelId: "kimi-k2.6",
+        streamFn: capturedStream.streamFn,
+      } as never),
+    ).toBe(capturedStream.streamFn);
+    expect(
+      provider.resolveThinkingProfile?.({
+        provider: "moonshot",
+        modelId: "kimi-k2.7-code",
+        reasoning: true,
+      } as never),
+    ).toEqual({
+      levels: [{ id: "low", label: "on" }],
+      defaultLevel: "low",
+      preserveWhenCatalogReasoningFalse: true,
+    });
+    expect(
+      provider.isModernModelRef?.({
+        provider: "moonshot",
+        modelId: "kimi-k2.7-code",
+      }),
+    ).toBe(true);
+    expect(
+      provider.isModernModelRef?.({
+        provider: "moonshot",
+        modelId: "kimi-k2.6",
+      }),
+    ).toBe(false);
+  });
 });

extensions/moonshot/index.ts

@@ -10,9 +10,11 @@ import {
   MOONSHOT_DEFAULT_MODEL_REF,
 } from "./onboard.js";
 import { buildMoonshotProvider } from "./provider-catalog.js";
+import { KIMI_K2_7_CODE_MODEL_ID, resolveThinkingProfile } from "./provider-policy-api.js";
 import { createKimiWebSearchProvider } from "./src/kimi-web-search-provider.js";
 
 const PROVIDER_ID = "moonshot";
+const moonshotThinkingStreamHooks = MOONSHOT_THINKING_STREAM_HOOKS;
 
 export default defineSingleProviderPluginEntry({
   id: PROVIDER_ID,
@@ -67,14 +69,13 @@ export default defineSingleProviderPluginEntry({
       sanitizeToolCallIds: false,
       dropReasoningFromHistory: false,
     }),
-    ...MOONSHOT_THINKING_STREAM_HOOKS,
-    resolveThinkingProfile: () => ({
-      levels: [
-        { id: "off", label: "off" },
-        { id: "low", label: "on" },
-      ],
-      defaultLevel: "off",
-    }),
+    ...moonshotThinkingStreamHooks,
+    wrapSimpleCompletionStreamFn: (ctx) =>
+      ctx.modelId.trim().toLowerCase() === KIMI_K2_7_CODE_MODEL_ID
+        ? moonshotThinkingStreamHooks.wrapStreamFn?.(ctx)
+        : ctx.streamFn,
+    resolveThinkingProfile,
+    isModernModelRef: ({ modelId }) => modelId.trim().toLowerCase() === KIMI_K2_7_CODE_MODEL_ID,
   },
   register(api) {
     api.registerMediaUnderstandingProvider(moonshotMediaUnderstandingProvider);

extensions/moonshot/moonshot.live.test.ts

@@ -1,11 +1,25 @@
 // Moonshot tests cover moonshot plugin behavior.
+import {
+  streamSimple,
+  type AssistantMessage,
+  type Context,
+  type Model,
+  type Tool,
+} from "openclaw/plugin-sdk/llm";
+import { registerSingleProviderPlugin } from "openclaw/plugin-sdk/plugin-test-runtime";
 import { isLiveTestEnabled } from "openclaw/plugin-sdk/test-env";
+import { Type } from "typebox";
 import { describe, expect, it } from "vitest";
+import plugin from "./index.js";
+import { buildMoonshotProvider, MOONSHOT_CN_BASE_URL } from "./provider-catalog.js";
 import { createKimiWebSearchProvider } from "./src/kimi-web-search-provider.js";
 
 const KIMI_SEARCH_KEY =
   process.env.KIMI_API_KEY?.trim() || process.env.MOONSHOT_API_KEY?.trim() || "";
+const MOONSHOT_API_KEY = process.env.MOONSHOT_API_KEY?.trim() || "";
 const describeLive = isLiveTestEnabled() && KIMI_SEARCH_KEY.length > 0 ? describe : describe.skip;
+const describeModelLive =
+  isLiveTestEnabled() && MOONSHOT_API_KEY.length > 0 ? describe : describe.skip;
 const KIMI_LIVE_SEARCH_TIMEOUT_SECONDS = 60;
 
 function isTransientKimiSearchError(error: unknown): boolean {
@@ -19,17 +33,31 @@ function isTransientKimiSearchError(error: unknown): boolean {
   return message.includes("timeout") || message.includes("aborted");
 }
 
-function isKimiAuthDrift(error: unknown): boolean {
+function isMoonshotAuthDrift(error: unknown): boolean {
   if (!(error instanceof Error)) {
     return false;
   }
   const message = error.message.toLowerCase();
   return (
-    message.includes("kimi api error (401)") &&
-    (message.includes("incorrect api key") || message.includes("incorrect_api_key"))
+    message.includes("401") &&
+    (message.includes("incorrect api key") ||
+      message.includes("incorrect_api_key") ||
+      message.includes("invalid authentication") ||
+      message.includes("invalid_authentication_error"))
   );
 }
 
+describe("moonshot live auth drift detection", () => {
+  it.each([
+    ["401 Incorrect API key provided", true],
+    ["401 invalid_authentication_error: Invalid Authentication", true],
+    ["401 Permission denied", false],
+    ["400 Incorrect API key provided", false],
+  ])("classifies %s", (message, expected) => {
+    expect(isMoonshotAuthDrift(new Error(message))).toBe(expected);
+  });
+});
+
 describeLive("moonshot plugin live", () => {
   it("runs Kimi web search through the provider tool", async () => {
     const provider = createKimiWebSearchProvider();
@@ -51,7 +79,7 @@ describeLive("moonshot plugin live", () => {
         break;
       } catch (error) {
         lastError = error;
-        if (isKimiAuthDrift(error)) {
+        if (isMoonshotAuthDrift(error)) {
           console.warn("[moonshot:live] skip Kimi web search: auth drift");
           return;
         }
@@ -71,6 +99,256 @@ describeLive("moonshot plugin live", () => {
   }, 180_000);
 });
 
+function resolveMoonshotModels(modelId: string): Model<"openai-completions">[] {
+  const provider = buildMoonshotProvider();
+  const model = provider.models.find((entry) => entry.id === modelId);
+  if (!model) {
+    throw new Error(`Moonshot catalog does not include ${modelId}`);
+  }
+  const defaultModel = {
+    provider: "moonshot",
+    baseUrl: provider.baseUrl,
+    ...model,
+    api: "openai-completions",
+  } as Model<"openai-completions">;
+  return [defaultModel, { ...defaultModel, baseUrl: MOONSHOT_CN_BASE_URL }];
+}
+
+function createNoopTool(): Tool {
+  return {
+    name: "noop",
+    description: "Return ok.",
+    parameters: Type.Object({}, { additionalProperties: false }),
+  };
+}
+
+async function collectDoneMessage(
+  stream: AsyncIterable<{ type: string; message?: AssistantMessage; error?: AssistantMessage }>,
+): Promise<AssistantMessage> {
+  let doneMessage: AssistantMessage | undefined;
+  for await (const event of stream) {
+    if (event.type === "error") {
+      throw new Error(event.error?.errorMessage || "Moonshot live request failed");
+    }
+    if (event.type === "done") {
+      doneMessage = event.message;
+    }
+  }
+  if (!doneMessage) {
+    throw new Error("Moonshot live stream ended without a done message");
+  }
+  return doneMessage;
+}
+
+describeModelLive("moonshot K2.6 replay live", () => {
+  it("accepts a cross-model tool-call replay after backfilling reasoning_content", async () => {
+    const provider = await registerSingleProviderPlugin(plugin);
+    const wrappedStream = provider.wrapStreamFn?.({
+      provider: "moonshot",
+      modelId: "kimi-k2.6",
+      thinkingLevel: "low",
+      streamFn: streamSimple,
+    } as never);
+    if (!wrappedStream) {
+      throw new Error("Moonshot provider did not register a stream wrapper");
+    }
+
+    const tool = createNoopTool();
+    const replayContext: Context = {
+      messages: [
+        {
+          role: "user",
+          content: "Call the noop tool.",
+          timestamp: Date.now(),
+        },
+        {
+          role: "assistant",
+          api: "openai-responses",
+          provider: "openai",
+          model: "gpt-5.5",
+          stopReason: "toolUse",
+          content: [{ type: "toolCall", id: "call_cross_model", name: "noop", arguments: {} }],
+          timestamp: Date.now(),
+        } as AssistantMessage,
+        {
+          role: "toolResult",
+          toolCallId: "call_cross_model",
+          toolName: "noop",
+          content: [{ type: "text", text: "ok" }],
+          isError: false,
+          timestamp: Date.now(),
+        },
+        {
+          role: "user",
+          content: "The tool returned ok. Reply with exactly: ok",
+          timestamp: Date.now(),
+        },
+      ],
+      tools: [tool],
+    };
+
+    const runScenario = async (model: Model<"openai-completions">) => {
+      let payload: Record<string, unknown> | undefined;
+      const response = await collectDoneMessage(
+        wrappedStream(model, replayContext, {
+          apiKey: MOONSHOT_API_KEY,
+          maxTokens: 256,
+          onPayload: (value) => {
+            payload = value as Record<string, unknown>;
+          },
+        }) as AsyncIterable<{
+          type: string;
+          message?: AssistantMessage;
+          error?: AssistantMessage;
+        }>,
+      );
+
+      const messages = payload?.messages as Array<Record<string, unknown>> | undefined;
+      const replayedAssistant = messages?.find(
+        (message) => message.role === "assistant" && Array.isArray(message.tool_calls),
+      );
+      expect(replayedAssistant?.reasoning_content).toBe("");
+      expect(response.stopReason).not.toBe("error");
+    };
+
+    let lastAuthError: unknown;
+    for (const model of resolveMoonshotModels("kimi-k2.6")) {
+      try {
+        await runScenario(model);
+        return;
+      } catch (error) {
+        if (!isMoonshotAuthDrift(error)) {
+          throw error;
+        }
+        lastAuthError = error;
+      }
+    }
+    throw toLintErrorObject(lastAuthError, "Moonshot K2.6 rejected the API key in both regions");
+  }, 180_000);
+});
+
+describeModelLive("moonshot K2.7 Code live", () => {
+  it("omits thinking controls and completes a replayed tool turn", async () => {
+    const provider = await registerSingleProviderPlugin(plugin);
+    const wrappedStream = provider.wrapStreamFn?.({
+      provider: "moonshot",
+      modelId: "kimi-k2.7-code",
+      thinkingLevel: "off",
+      extraParams: { thinking: { type: "disabled", keep: "all" } },
+      streamFn: streamSimple,
+    } as never);
+    if (!wrappedStream) {
+      throw new Error("Moonshot provider did not register a stream wrapper");
+    }
+
+    const tool = createNoopTool();
+    const firstUser = {
+      role: "user" as const,
+      content: "Call the noop tool with {}. Do not answer directly.",
+      timestamp: Date.now(),
+    };
+
+    const runScenario = async (model: Model<"openai-completions">) => {
+      let firstPayload: Record<string, unknown> | undefined;
+      const first = await collectDoneMessage(
+        wrappedStream(
+          model,
+          { messages: [firstUser], tools: [tool] },
+          {
+            apiKey: MOONSHOT_API_KEY,
+            maxTokens: 16_000,
+            temperature: 0,
+            onPayload: (payload) => {
+              firstPayload = payload as Record<string, unknown>;
+            },
+          },
+        ) as AsyncIterable<{
+          type: string;
+          message?: AssistantMessage;
+          error?: AssistantMessage;
+        }>,
+      );
+
+      expect(firstPayload).toBeDefined();
+      expect(firstPayload).not.toHaveProperty("thinking");
+      expect(firstPayload).not.toHaveProperty("reasoning_effort");
+      expect(firstPayload).not.toHaveProperty("temperature");
+      const reasoning = first.content.find((block) => block.type === "thinking");
+      if (!reasoning || reasoning.type !== "thinking" || reasoning.thinking.length === 0) {
+        throw new Error("Moonshot K2.7 Code did not return captured reasoning");
+      }
+      const toolCall = first.content.find((block) => block.type === "toolCall");
+      if (!toolCall || toolCall.type !== "toolCall") {
+        throw new Error(`Moonshot K2.7 Code did not call noop: ${first.stopReason}`);
+      }
+      expect(toolCall.name).toBe("noop");
+
+      let secondPayload: Record<string, unknown> | undefined;
+      const replayContext: Context = {
+        messages: [
+          firstUser,
+          first,
+          {
+            role: "toolResult",
+            toolCallId: toolCall.id,
+            toolName: toolCall.name,
+            content: [{ type: "text", text: "ok" }],
+            isError: false,
+            timestamp: Date.now(),
+          },
+          {
+            role: "user",
+            content: "Reply with exactly: ok",
+            timestamp: Date.now(),
+          },
+        ],
+        tools: [tool],
+      };
+      const second = await collectDoneMessage(
+        wrappedStream(model, replayContext, {
+          apiKey: MOONSHOT_API_KEY,
+          maxTokens: 16_000,
+          temperature: 0,
+          onPayload: (payload) => {
+            secondPayload = payload as Record<string, unknown>;
+          },
+        }) as AsyncIterable<{
+          type: string;
+          message?: AssistantMessage;
+          error?: AssistantMessage;
+        }>,
+      );
+
+      expect(secondPayload).toBeDefined();
+      expect(secondPayload).not.toHaveProperty("thinking");
+      expect(secondPayload).not.toHaveProperty("reasoning_effort");
+      expect(secondPayload).not.toHaveProperty("temperature");
+      const text = second.content
+        .filter((block) => block.type === "text")
+        .map((block) => block.text.trim())
+        .join(" ");
+      expect(text).toMatch(/^ok[.!]?$/i);
+    };
+
+    let lastAuthError: unknown;
+    for (const model of resolveMoonshotModels("kimi-k2.7-code")) {
+      try {
+        await runScenario(model);
+        return;
+      } catch (error) {
+        if (!isMoonshotAuthDrift(error)) {
+          throw error;
+        }
+        lastAuthError = error;
+      }
+    }
+    throw toLintErrorObject(
+      lastAuthError,
+      "Moonshot K2.7 Code rejected the API key in both regions",
+    );
+  }, 180_000);
+});
+
 function toLintErrorObject(value: unknown, fallbackMessage: string): Error {
   if (value instanceof Error) {
     return value;

extensions/moonshot/openclaw.plugin.json

@@ -63,6 +63,20 @@
               "cacheWrite": 0
             }
           },
+          {
+            "id": "kimi-k2.7-code",
+            "name": "Kimi K2.7 Code",
+            "reasoning": true,
+            "input": ["text", "image"],
+            "contextWindow": 262144,
+            "maxTokens": 262144,
+            "cost": {
+              "input": 0.95,
+              "output": 4,
+              "cacheRead": 0.19,
+              "cacheWrite": 0
+            }
+          },
           {
             "id": "kimi-k2.5",
             "name": "Kimi K2.5",

extensions/moonshot/provider-catalog.test.ts

@@ -41,6 +41,7 @@ describe("moonshot provider catalog", () => {
     expect(provider.api).toBe("openai-completions");
     expect(provider.models.map((model) => model.id)).toEqual([
       "kimi-k2.6",
+      "kimi-k2.7-code",
       "kimi-k2.5",
       "kimi-k2-thinking",
       "kimi-k2-thinking-turbo",
@@ -52,6 +53,18 @@ describe("moonshot provider catalog", () => {
       cacheRead: 0.16,
       cacheWrite: 0,
     });
+    expect(requireMoonshotModel(provider, "kimi-k2.7-code")).toMatchObject({
+      reasoning: true,
+      input: ["text", "image"],
+      contextWindow: 262144,
+      maxTokens: 262144,
+      cost: {
+        input: 0.95,
+        output: 4,
+        cacheRead: 0.19,
+        cacheWrite: 0,
+      },
+    });
     expect(requireMoonshotModel(provider, "kimi-k2.5").cost).toEqual({
       input: 0.6,
       output: 3,

extensions/moonshot/provider-policy-api.ts

@@ -0,0 +1,21 @@
+// Moonshot policy module exposes model-specific thinking controls before runtime registration.
+import type { ProviderDefaultThinkingPolicyContext } from "openclaw/plugin-sdk/core";
+
+export const KIMI_K2_7_CODE_MODEL_ID = "kimi-k2.7-code";
+
+export function resolveThinkingProfile(context: ProviderDefaultThinkingPolicyContext) {
+  if (context.modelId.trim().toLowerCase() === KIMI_K2_7_CODE_MODEL_ID) {
+    return {
+      levels: [{ id: "low" as const, label: "on" }],
+      defaultLevel: "low" as const,
+      preserveWhenCatalogReasoningFalse: true,
+    };
+  }
+  return {
+    levels: [
+      { id: "off" as const, label: "off" },
+      { id: "low" as const, label: "on" },
+    ],
+    defaultLevel: "off" as const,
+  };
+}

extensions/qa-lab/src/cli.runtime.test.ts

@@ -82,6 +82,8 @@ import {
   runQaParityReportCommand,
   runQaSuiteCommand,
 } from "./cli.runtime.js";
+import { QaSuiteInfraError } from "./errors.js";
+import { QA_EVIDENCE_FILENAME } from "./evidence-summary.js";
 import { runQaTelegramCommand } from "./live-transports/telegram/cli.runtime.js";
 import { defaultQaModelForMode as defaultQaProviderModelForMode } from "./model-selection.js";
 import type { QaProviderModeInput } from "./run-config.js";
@@ -127,7 +129,7 @@ describe("qa cli runtime", () => {
     suiteReportPath = path.join(suiteArtifactsDir, "qa-suite-report.md");
     suiteSummaryPath = path.join(suiteArtifactsDir, "qa-suite-summary.json");
     telegramArtifactsDir = await fs.mkdtemp(path.join(os.tmpdir(), "qa-telegram-runtime-"));
-    telegramSummaryPath = path.join(telegramArtifactsDir, "telegram-qa-summary.json");
+    telegramSummaryPath = path.join(telegramArtifactsDir, QA_EVIDENCE_FILENAME);
     await fs.writeFile(suiteReportPath, "# QA Suite Report\n", "utf8");
     await fs.writeFile(
       suiteSummaryPath,
@@ -616,7 +618,9 @@ describe("qa cli runtime", () => {
 
   it("retries host suite runs once for retryable infra failures", async () => {
     runQaSuiteFromRuntime
-      .mockRejectedValueOnce(new Error("agent.wait timeout while waiting for transport ready"))
+      .mockRejectedValueOnce(
+        new QaSuiteInfraError("agent_wait_failed", "agent.wait failed: gateway call timed out"),
+      )
       .mockResolvedValueOnce({
         watchUrl: "http://127.0.0.1:43124",
         reportPath: suiteReportPath,
@@ -629,13 +633,14 @@ describe("qa cli runtime", () => {
     });
 
     expect(runQaSuiteFromRuntime).toHaveBeenCalledTimes(2);
-    expectWriteContains(stderrWrite, "[qa-suite] infra retry 1/1: agent.wait timeout");
+    expectWriteContains(stderrWrite, "[qa-suite] infra retry 1/1: agent.wait failed");
   });
 
   it("retries host suite runs once for qa-channel readiness timeouts", async () => {
     runQaSuiteFromRuntime
       .mockRejectedValueOnce(
-        new Error(
+        new QaSuiteInfraError(
+          "transport_ready_timeout",
           "timed out after 180000ms waiting for qa-channel ready; last status: no qa-channel accounts reported",
         ),
       )
@@ -1658,7 +1663,9 @@ describe("qa cli runtime", () => {
           repoRoot: "/tmp/openclaw-repo",
           runner: "multipass",
         }),
-      ).rejects.toThrow("did not include counts.failed, counts.skipped, or scenarios[].status");
+      ).rejects.toThrow(
+        "did not include counts.failed, counts.skipped, scenarios[].status, or entries[].result.status",
+      );
     } finally {
       await fs.rm(repoRoot, { recursive: true, force: true });
     }

extensions/qa-lab/src/cli.runtime.ts

@@ -28,6 +28,7 @@ import {
 } from "./coverage-report.js";
 import { buildQaDockerHarnessImage, writeQaDockerHarnessFiles } from "./docker-harness.js";
 import { runQaDockerUp } from "./docker-up.runtime.js";
+import { QaSuiteArtifactError, QaSuiteInfraError } from "./errors.js";
 import type { QaCliBackendAuthMode } from "./gateway-child.js";
 import {
   createMockJsonlReplayCellRunner,
@@ -81,6 +82,13 @@ import {
 } from "./tool-coverage-report.js";
 
 const QA_SUITE_INFRA_RETRY_LIMIT = 1;
+const QA_SUITE_INFRA_RETRY_NETWORK_ERROR_CODES = new Set([
+  "ECONNRESET",
+  "ECONNREFUSED",
+  "EPIPE",
+  "ETIMEDOUT",
+  "UND_ERR_SOCKET",
+]);
 
 type InterruptibleServer = {
   baseUrl: string;
@@ -244,29 +252,36 @@ function resolveQaRuntimeParityTierScenarioIds(params: {
 }
 
 function isQaSuiteInfraRetryableError(error: unknown) {
-  const message = formatErrorMessage(error).toLowerCase();
-  return (
-    message.includes("agent.wait timeout") ||
-    message.includes("qa cli timed out") ||
-    message.includes("readyz") ||
-    message.includes("gateway healthy") ||
-    message.includes("transport ready") ||
-    message.includes("waiting for qa-channel ready") ||
-    message.includes("econnreset") ||
-    message.includes("econnrefused") ||
-    message.includes("socket hang up") ||
-    message.includes("could not read qa summary json") ||
-    message.includes("could not parse qa summary json") ||
-    message.includes("did not include counts.failed, counts.skipped, or scenarios[].status") ||
-    message.includes("did not produce report artifact")
-  );
+  if (error instanceof QaSuiteArtifactError || error instanceof QaSuiteInfraError) {
+    return true;
+  }
+  return hasQaSuiteRetryableNetworkCode(error);
+}
+
+function hasQaSuiteRetryableNetworkCode(error: unknown) {
+  let current: unknown = error;
+  for (let depth = 0; depth < 4 && current; depth += 1) {
+    if (typeof current !== "object") {
+      return false;
+    }
+    const record = current as { cause?: unknown; code?: unknown };
+    if (
+      typeof record.code === "string" &&
+      QA_SUITE_INFRA_RETRY_NETWORK_ERROR_CODES.has(record.code.toUpperCase())
+    ) {
+      return true;
+    }
+    current = record.cause;
+  }
+  return false;
 }
 
 async function assertQaSuiteArtifacts(result: { reportPath: string; summaryPath: string }) {
   try {
     await fs.access(result.reportPath);
   } catch (error) {
-    throw new Error(
+    throw new QaSuiteArtifactError(
+      "report_missing",
       `QA suite did not produce report artifact at ${result.reportPath}: ${formatErrorMessage(error)}`,
       { cause: error },
     );

extensions/qa-lab/src/coverage-report.test.ts

@@ -2,6 +2,31 @@
 import { describe, expect, it } from "vitest";
 import { buildQaCoverageInventory, renderQaCoverageMarkdownReport } from "./coverage-report.js";
 import { readQaScenarioPack } from "./scenario-catalog.js";
+import { buildQaScorecardTaxonomyReport, parseQaScorecardTaxonomy } from "./scorecard-taxonomy.js";
+
+const TEST_EXECUTABLE_CATEGORY_ID = "agent-runtime-and-provider-execution.agent-turn-execution";
+const TEST_TAXONOMY_REF = {
+  sourcePath: "taxonomy.yaml",
+  version: 1,
+  processVersion: 3,
+  snapshotDate: "2026-05-26",
+  sourceRef: "origin/main@41eef4a7965",
+};
+
+function testScorecardProfiles(categoryId = TEST_EXECUTABLE_CATEGORY_ID, profileId = "release") {
+  return [
+    {
+      id: "smoke-ci",
+      description: "Test smoke profile.",
+      categoryIds: profileId === "smoke-ci" ? [categoryId] : [],
+    },
+    {
+      id: "release",
+      description: "Test release profile.",
+      categoryIds: profileId === "release" ? [categoryId] : [],
+    },
+  ];
+}
 
 describe("qa coverage report", () => {
   it("groups scenario coverage metadata by theme and surface", () => {
@@ -19,6 +44,43 @@ describe("qa coverage report", () => {
       "telegram",
       "whatsapp",
     ]);
+    expect(inventory.scorecardTaxonomy.taxonomyId).toBe("stable-lts-initial");
+    expect(inventory.scorecardTaxonomy.profileCount).toBe(2);
+    expect(inventory.scorecardTaxonomy.categoryCount).toBe(16);
+    expect(inventory.scorecardTaxonomy.ltsIncludedCategoryCount).toBe(7);
+    expect(inventory.scorecardTaxonomy.deferredCategoryCount).toBe(8);
+    expect(inventory.scorecardTaxonomy.advisoryCategoryCount).toBe(1);
+    expect(inventory.scorecardTaxonomy.releaseBlockingCategoryCount).toBe(7);
+    expect(inventory.scorecardTaxonomy.mappedCoverageIdCount).toBeGreaterThan(0);
+    expect(inventory.scorecardTaxonomy.mappedScenarioCount).toBeGreaterThan(0);
+    expect(inventory.scorecardTaxonomy.unmappedCoverageIdCount).toBeGreaterThan(0);
+    expect(inventory.scorecardTaxonomy.validationIssues).toStrictEqual([]);
+    expect(
+      inventory.scorecardTaxonomy.profiles
+        .find((profile) => profile.id === "release")
+        ?.categoryIds.toSorted(),
+    ).toEqual([
+      "agent-runtime-and-provider-execution.agent-turn-execution",
+      "automation-cron-hooks-tasks-polling.cron-jobs",
+      "browser-automation-and-exec-sandbox-tools.tool-invocation-and-execution",
+      "browser-control-ui-and-webchat.browser-ui",
+      "media-understanding-and-media-generation.media-generation",
+      "media-understanding-and-media-generation.media-understanding",
+      "openai-codex-provider-path.responses-and-tool-compatibility",
+      "plugin-sdk-and-bundled-plugin-architecture.installing-and-running-plugins",
+      "security-auth-pairing-and-secrets.approval-policy-and-tool-safeguards",
+      "security-auth-pairing-and-secrets.credential-and-secret-hygiene",
+      "session-memory-and-context-engine.diagnostics-maintenance-and-recovery",
+      "session-memory-and-context-engine.memory",
+      "session-memory-and-context-engine.token-management",
+      "telemetry-diagnostics-and-observability.telemetry-export",
+    ]);
+    expect(
+      inventory.scorecardTaxonomy.categories.find(
+        (category) =>
+          category.id === "clawhub-and-external-plugin-distribution.compatibility-and-trust",
+      )?.profiles,
+    ).toStrictEqual([]);
     expect(inventory.scenarioPacks.map((pack) => pack.id)).toEqual([
       "observability",
       "personal-agent",
@@ -60,5 +122,414 @@ describe("qa coverage report", () => {
       "- telegram (telegram): canary: always-on, help-command: telegram-help-command, mention-gating: telegram-mention-gating; missing baseline: allowlist-block, top-level-reply-shape, restart-resume",
     );
     expect(report).toContain("thread-follow-up: slack-thread-follow-up");
+    expect(report).toContain("## Scorecard Taxonomy");
+    expect(report).toContain("- Mapping ID: stable-lts-initial");
+    expect(report).toContain("- Maturity taxonomy: taxonomy.yaml");
+    expect(report).toContain("- Maturity score snapshot: docs/maturity-scores.yaml");
+    expect(report).toContain("- Categories: 16 (7 LTS-included, 8 deferred, 1 advisory)");
+    expect(report).toContain("- Profiles: 2");
+    expect(report).toContain(
+      "- smoke-ci: 14 categories; agent-runtime-and-provider-execution.agent-turn-execution,",
+    );
+    expect(report).toContain(
+      "- browser-automation-and-exec-sandbox-tools.tool-invocation-and-execution (browser-automation-and-exec-sandbox-tools / Tool Invocation and Execution; lts-included, release-blocking, mapped): profiles: release, smoke-ci; coverage: tools.apply-patch, tools.exec, tools.fs.read, tools.fs.write, tools.web-search;",
+    );
+    expect(report).toContain("### Unmapped Coverage IDs");
+    expect(report).toContain("agents.subagents");
+  });
+
+  it("reports taxonomy mapping gaps as scorecard signals", () => {
+    const taxonomy = parseQaScorecardTaxonomy({
+      version: 1,
+      id: "test-taxonomy",
+      title: "Test taxonomy",
+      taxonomy: TEST_TAXONOMY_REF,
+      scoreSnapshotRef: "docs/maturity-scores.yaml",
+      status: "initial",
+      profiles: testScorecardProfiles(),
+      categories: [
+        {
+          id: TEST_EXECUTABLE_CATEGORY_ID,
+          taxonomySurfaceId: "agent-runtime-and-provider-execution",
+          taxonomyCategoryName: "Agent Turn Execution",
+          supportStatus: "lts-included",
+          releaseBlocking: true,
+          requirement: "Exercise a missing mapping.",
+          evidenceRequired: "A real scenario mapping before promotion.",
+          evidence: {
+            profiles: ["release"],
+            liveProofRequired: false,
+            freshness: "target-ref",
+            coverageIds: ["runtime.missing-coverage"],
+            scenarioRefs: ["qa/scenarios/runtime/missing-scorecard-scenario.md"],
+            docsRefs: ["docs/missing-scorecard-doc.md"],
+            codeRefs: ["src/missing-scorecard-code.ts"],
+          },
+        },
+      ],
+    });
+
+    const report = buildQaScorecardTaxonomyReport({
+      taxonomy,
+      repoRoot: process.cwd(),
+      scenarios: readQaScenarioPack().scenarios,
+    });
+
+    expect(report.categories[0]?.mappingStatus).toBe("partial");
+    expect(report.validationIssues.map((issue) => issue.code)).toEqual([
+      "coverage-id-not-found",
+      "scenario-ref-not-found",
+      "docs-ref-not-found",
+      "code-ref-not-found",
+    ]);
+  });
+
+  it("reports release-blocking categories missing release profile membership", () => {
+    const taxonomy = parseQaScorecardTaxonomy({
+      version: 1,
+      id: "test-taxonomy",
+      title: "Test taxonomy",
+      taxonomy: TEST_TAXONOMY_REF,
+      scoreSnapshotRef: "docs/maturity-scores.yaml",
+      status: "initial",
+      profiles: testScorecardProfiles(TEST_EXECUTABLE_CATEGORY_ID, "smoke-ci"),
+      categories: [
+        {
+          id: TEST_EXECUTABLE_CATEGORY_ID,
+          taxonomySurfaceId: "agent-runtime-and-provider-execution",
+          taxonomyCategoryName: "Agent Turn Execution",
+          supportStatus: "lts-included",
+          releaseBlocking: true,
+          requirement: "Release-blocking rows must be selected by the release profile.",
+          evidenceRequired: "Release profile membership before promotion.",
+          evidence: {
+            profiles: ["smoke-ci"],
+            liveProofRequired: false,
+            freshness: "target-ref",
+            coverageIds: ["channels.dm"],
+            scenarioRefs: ["qa/scenarios/channels/dm-chat-baseline.md"],
+            docsRefs: ["docs/concepts/qa-e2e-automation.md"],
+            codeRefs: ["extensions/qa-lab/src/suite.ts"],
+          },
+        },
+      ],
+    });
+
+    const report = buildQaScorecardTaxonomyReport({
+      taxonomy,
+      repoRoot: process.cwd(),
+      scenarios: readQaScenarioPack().scenarios,
+    });
+
+    expect(report.validationIssues.map((issue) => issue.code)).toEqual([
+      "release-blocking-category-missing-release-profile",
+    ]);
+  });
+
+  it("reports advisory categories that are accidentally assigned to a runnable profile", () => {
+    const taxonomy = parseQaScorecardTaxonomy({
+      version: 1,
+      id: "test-taxonomy",
+      title: "Test taxonomy",
+      taxonomy: TEST_TAXONOMY_REF,
+      scoreSnapshotRef: "docs/maturity-scores.yaml",
+      status: "initial",
+      profiles: testScorecardProfiles(
+        "clawhub-and-external-plugin-distribution.compatibility-and-trust",
+        "smoke-ci",
+      ),
+      categories: [
+        {
+          id: "clawhub-and-external-plugin-distribution.compatibility-and-trust",
+          taxonomySurfaceId: "clawhub-and-external-plugin-distribution",
+          taxonomyCategoryName: "Compatibility and Trust",
+          supportStatus: "advisory",
+          releaseBlocking: false,
+          requirement: "Keep advisory compatibility out of runnable profiles.",
+          evidenceRequired: "Advisory report metadata only.",
+          evidence: {
+            profiles: [],
+            liveProofRequired: false,
+            freshness: "latest-advisory-run",
+            coverageIds: [],
+            scenarioRefs: [],
+            docsRefs: ["docs/plugins/architecture.md"],
+            codeRefs: [],
+          },
+        },
+      ],
+    });
+
+    const report = buildQaScorecardTaxonomyReport({
+      taxonomy,
+      repoRoot: process.cwd(),
+      scenarios: readQaScenarioPack().scenarios,
+    });
+
+    expect(report.validationIssues.map((issue) => issue.code)).toEqual([
+      "profile-membership-missing-category-profile",
+      "advisory-category-has-profile-membership",
+    ]);
+  });
+
+  it("reports non-advisory categories with no runnable profile membership", () => {
+    const taxonomy = parseQaScorecardTaxonomy({
+      version: 1,
+      id: "test-taxonomy",
+      title: "Test taxonomy",
+      taxonomy: TEST_TAXONOMY_REF,
+      scoreSnapshotRef: "docs/maturity-scores.yaml",
+      status: "initial",
+      profiles: testScorecardProfiles(TEST_EXECUTABLE_CATEGORY_ID, "none"),
+      categories: [
+        {
+          id: TEST_EXECUTABLE_CATEGORY_ID,
+          taxonomySurfaceId: "agent-runtime-and-provider-execution",
+          taxonomyCategoryName: "Agent Turn Execution",
+          supportStatus: "deferred",
+          releaseBlocking: false,
+          requirement: "Non-advisory rows must stay visible to runnable profiles.",
+          evidenceRequired: "At least one smoke-ci or release membership before promotion.",
+          evidence: {
+            profiles: [],
+            liveProofRequired: false,
+            freshness: "target-ref",
+            coverageIds: ["channels.dm"],
+            scenarioRefs: ["qa/scenarios/channels/dm-chat-baseline.md"],
+            docsRefs: ["docs/concepts/qa-e2e-automation.md"],
+            codeRefs: ["extensions/qa-lab/src/suite.ts"],
+          },
+        },
+      ],
+    });
+
+    const report = buildQaScorecardTaxonomyReport({
+      taxonomy,
+      repoRoot: process.cwd(),
+      scenarios: readQaScenarioPack().scenarios,
+    });
+
+    expect(report.validationIssues.map((issue) => issue.code)).toEqual([
+      "non-advisory-category-missing-profile-membership",
+    ]);
+  });
+
+  it("reports executable category refs missing from taxonomy.yaml", () => {
+    const taxonomy = parseQaScorecardTaxonomy({
+      version: 1,
+      id: "test-taxonomy",
+      title: "Test taxonomy",
+      taxonomy: TEST_TAXONOMY_REF,
+      scoreSnapshotRef: "docs/maturity-scores.yaml",
+      status: "initial",
+      profiles: testScorecardProfiles(TEST_EXECUTABLE_CATEGORY_ID, "release"),
+      categories: [
+        {
+          id: TEST_EXECUTABLE_CATEGORY_ID,
+          taxonomySurfaceId: "agent-runtime-and-provider-execution",
+          taxonomyCategoryName: "Missing Taxonomy Category",
+          supportStatus: "lts-included",
+          releaseBlocking: true,
+          requirement: "Executable refs must resolve against taxonomy.yaml.",
+          evidenceRequired: "A valid taxonomy surface/category ref.",
+          evidence: {
+            profiles: ["release"],
+            liveProofRequired: false,
+            freshness: "target-ref",
+            coverageIds: ["channels.dm"],
+            scenarioRefs: ["qa/scenarios/channels/dm-chat-baseline.md"],
+            docsRefs: ["docs/concepts/qa-e2e-automation.md"],
+            codeRefs: ["extensions/qa-lab/src/suite.ts"],
+          },
+        },
+      ],
+    });
+
+    const report = buildQaScorecardTaxonomyReport({
+      taxonomy,
+      repoRoot: process.cwd(),
+      scenarios: readQaScenarioPack().scenarios,
+    });
+
+    expect(report.validationIssues.map((issue) => issue.code)).toEqual([
+      "taxonomy-category-ref-not-found",
+    ]);
+  });
+
+  it("reports profile membership refs missing from executable categories", () => {
+    const taxonomy = parseQaScorecardTaxonomy({
+      version: 1,
+      id: "test-taxonomy",
+      title: "Test taxonomy",
+      taxonomy: TEST_TAXONOMY_REF,
+      scoreSnapshotRef: "docs/maturity-scores.yaml",
+      status: "initial",
+      profiles: [
+        {
+          id: "smoke-ci",
+          description: "Test smoke profile.",
+          categoryIds: ["missing.category"],
+        },
+        {
+          id: "release",
+          description: "Test release profile.",
+          categoryIds: [],
+        },
+      ],
+      categories: [
+        {
+          id: TEST_EXECUTABLE_CATEGORY_ID,
+          taxonomySurfaceId: "agent-runtime-and-provider-execution",
+          taxonomyCategoryName: "Agent Turn Execution",
+          supportStatus: "advisory",
+          releaseBlocking: false,
+          requirement: "Profile selectors must reference executable category IDs.",
+          evidenceRequired: "Invalid selector refs should be reported.",
+          evidence: {
+            profiles: [],
+            liveProofRequired: false,
+            freshness: "latest-advisory-run",
+            coverageIds: [],
+            scenarioRefs: [],
+            docsRefs: ["docs/concepts/qa-e2e-automation.md"],
+            codeRefs: ["extensions/qa-lab/src/suite.ts"],
+          },
+        },
+      ],
+    });
+
+    const report = buildQaScorecardTaxonomyReport({
+      taxonomy,
+      repoRoot: process.cwd(),
+      scenarios: readQaScenarioPack().scenarios,
+    });
+
+    expect(report.validationIssues.map((issue) => issue.code)).toEqual([
+      "profile-category-ref-not-found",
+    ]);
+  });
+
+  it("reports category profile refs missing from top-level mapping profiles", () => {
+    const taxonomy = parseQaScorecardTaxonomy({
+      version: 1,
+      id: "test-taxonomy",
+      title: "Test taxonomy",
+      taxonomy: TEST_TAXONOMY_REF,
+      scoreSnapshotRef: "docs/maturity-scores.yaml",
+      status: "initial",
+      profiles: [...testScorecardProfiles(TEST_EXECUTABLE_CATEGORY_ID, "release")],
+      categories: [
+        {
+          id: TEST_EXECUTABLE_CATEGORY_ID,
+          taxonomySurfaceId: "agent-runtime-and-provider-execution",
+          taxonomyCategoryName: "Agent Turn Execution",
+          supportStatus: "lts-included",
+          releaseBlocking: true,
+          requirement: "Category profile refs must resolve to top-level mapping profiles.",
+          evidenceRequired: "Unknown profile refs should be reported.",
+          evidence: {
+            profiles: ["release", "nightly"],
+            liveProofRequired: false,
+            freshness: "target-ref",
+            coverageIds: ["channels.dm"],
+            scenarioRefs: ["qa/scenarios/channels/dm-chat-baseline.md"],
+            docsRefs: ["docs/concepts/qa-e2e-automation.md"],
+            codeRefs: ["extensions/qa-lab/src/suite.ts"],
+          },
+        },
+      ],
+    });
+
+    const report = buildQaScorecardTaxonomyReport({
+      taxonomy,
+      repoRoot: process.cwd(),
+      scenarios: readQaScenarioPack().scenarios,
+    });
+
+    expect(report.validationIssues.map((issue) => issue.code)).toEqual(["profile-ref-not-found"]);
+  });
+
+  it("counts declared custom profiles as runnable category membership", () => {
+    const taxonomy = parseQaScorecardTaxonomy({
+      version: 1,
+      id: "test-taxonomy",
+      title: "Test taxonomy",
+      taxonomy: TEST_TAXONOMY_REF,
+      scoreSnapshotRef: "docs/maturity-scores.yaml",
+      status: "initial",
+      profiles: [
+        ...testScorecardProfiles(TEST_EXECUTABLE_CATEGORY_ID, "none"),
+        {
+          id: "nightly",
+          description: "Nightly mapped profile.",
+          categoryIds: [TEST_EXECUTABLE_CATEGORY_ID],
+        },
+      ],
+      categories: [
+        {
+          id: TEST_EXECUTABLE_CATEGORY_ID,
+          taxonomySurfaceId: "agent-runtime-and-provider-execution",
+          taxonomyCategoryName: "Agent Turn Execution",
+          supportStatus: "deferred",
+          releaseBlocking: false,
+          requirement: "Declared profile names can satisfy runnable coverage.",
+          evidenceRequired: "Profile names come from taxonomy-mappings.yaml.",
+          evidence: {
+            profiles: ["nightly"],
+            liveProofRequired: false,
+            freshness: "target-ref",
+            coverageIds: ["channels.dm"],
+            scenarioRefs: ["qa/scenarios/channels/dm-chat-baseline.md"],
+            docsRefs: ["docs/concepts/qa-e2e-automation.md"],
+            codeRefs: ["extensions/qa-lab/src/suite.ts"],
+          },
+        },
+      ],
+    });
+
+    const report = buildQaScorecardTaxonomyReport({
+      taxonomy,
+      repoRoot: process.cwd(),
+      scenarios: readQaScenarioPack().scenarios,
+    });
+
+    expect(report.validationIssues).toStrictEqual([]);
+  });
+
+  it("rejects taxonomy refs outside the repository", () => {
+    expect(() =>
+      parseQaScorecardTaxonomy({
+        version: 1,
+        id: "bad-taxonomy",
+        title: "Bad taxonomy",
+        taxonomy: {
+          ...TEST_TAXONOMY_REF,
+          sourcePath: "../rfcs/rfcs/0007-e2e-qa-lab-scorecard-consolidation.md",
+        },
+        scoreSnapshotRef: "docs/maturity-scores.yaml",
+        status: "initial",
+        profiles: testScorecardProfiles(TEST_EXECUTABLE_CATEGORY_ID, "smoke-ci"),
+        categories: [
+          {
+            id: TEST_EXECUTABLE_CATEGORY_ID,
+            taxonomySurfaceId: "agent-runtime-and-provider-execution",
+            taxonomyCategoryName: "Agent Turn Execution",
+            supportStatus: "deferred",
+            releaseBlocking: false,
+            requirement: "Reject escaped refs.",
+            evidenceRequired: "Parser rejects refs outside the repository.",
+            evidence: {
+              profiles: ["smoke-ci"],
+              liveProofRequired: false,
+              freshness: "target-ref",
+              coverageIds: ["runtime.delivery"],
+              scenarioRefs: ["qa/scenarios/channels/dm-chat-baseline.md"],
+              docsRefs: ["/tmp/outside-openclaw.md"],
+              codeRefs: ["src/agents/../agents/agent-tools.ts"],
+            },
+          },
+        ],
+      }),
+    ).toThrow("repo refs must not be absolute or contain parent-directory segments");
   });
 });

extensions/qa-lab/src/coverage-report.ts

@@ -5,6 +5,10 @@ import {
   type LiveTransportCoverageLaneSummary,
 } from "./live-transports/shared/live-transport-scenarios.js";
 import { QA_SCENARIO_PACKS, type QaSeedScenarioWithSource } from "./scenario-catalog.js";
+import {
+  readQaScorecardTaxonomyReport,
+  type QaScorecardTaxonomyReport,
+} from "./scorecard-taxonomy.js";
 
 type QaCoverageScenarioSummary = {
   id: string;
@@ -56,6 +60,7 @@ type QaCoverageInventory = {
   bySurface: Record<string, QaCoverageFeatureSummary[]>;
   scenarioPacks: QaCoverageScenarioPackSummary[];
   liveTransportLanes: LiveTransportCoverageLaneSummary[];
+  scorecardTaxonomy: QaScorecardTaxonomyReport;
 };
 
 function scenarioTheme(sourcePath: string) {
@@ -265,6 +270,7 @@ export function buildQaCoverageInventory(
     bySurface,
     scenarioPacks: buildScenarioPackSummaries(scenarios),
     liveTransportLanes: buildLiveTransportCoverageLaneSummaries(),
+    scorecardTaxonomy: readQaScorecardTaxonomyReport(scenarios),
   };
 }
 
@@ -310,6 +316,64 @@ function pushScenarioPackLines(lines: string[], packs: readonly QaCoverageScenar
   }
 }
 
+function pushScorecardTaxonomyLines(lines: string[], report: QaScorecardTaxonomyReport) {
+  lines.push("## Scorecard Taxonomy", "");
+  lines.push(`- Mapping: ${report.taxonomyPath ?? "missing"}`);
+  lines.push(`- Mapping ID: ${report.taxonomyId ?? "missing"}`);
+  lines.push(`- Maturity taxonomy: ${report.taxonomy?.sourcePath ?? "missing"}`);
+  if (report.scoreSnapshotRef) {
+    lines.push(`- Maturity score snapshot: ${report.scoreSnapshotRef}`);
+  }
+  lines.push(
+    `- Categories: ${report.categoryCount} (${report.ltsIncludedCategoryCount} LTS-included, ${report.deferredCategoryCount} deferred, ${report.advisoryCategoryCount} advisory)`,
+  );
+  lines.push(`- Profiles: ${report.profileCount}`);
+  lines.push(`- Release-blocking categories: ${report.releaseBlockingCategoryCount}`);
+  lines.push(`- Mapped coverage IDs: ${report.mappedCoverageIdCount}`);
+  lines.push(`- Mapped scenarios: ${report.mappedScenarioCount}`);
+  lines.push(`- Unmapped coverage IDs: ${report.unmappedCoverageIdCount}`);
+  lines.push(`- Validation warnings: ${report.validationIssueCount}`, "");
+
+  if (report.profiles.length > 0) {
+    lines.push("### Profiles", "");
+    for (const profile of report.profiles) {
+      const categories = profile.categoryIds.length > 0 ? profile.categoryIds.join(", ") : "none";
+      lines.push(`- ${profile.id}: ${profile.categoryIds.length} categories; ${categories}`);
+    }
+    lines.push("");
+  }
+
+  if (report.categories.length > 0) {
+    lines.push("### Category Mapping", "");
+    for (const category of report.categories) {
+      const blocking = category.releaseBlocking ? "release-blocking" : "non-blocking";
+      const coverage = category.coverageIds.length > 0 ? category.coverageIds.join(", ") : "none";
+      const scenarios =
+        category.scenarioRefs.length > 0 ? category.scenarioRefs.join(", ") : "none";
+      const profiles = category.profiles.length > 0 ? category.profiles.join(", ") : "none";
+      lines.push(
+        `- ${category.id} (${category.taxonomySurfaceId} / ${category.taxonomyCategoryName}; ${category.supportStatus}, ${blocking}, ${category.mappingStatus}): profiles: ${profiles}; coverage: ${coverage}; scenarios: ${scenarios}`,
+      );
+    }
+    lines.push("");
+  }
+
+  if (report.validationIssues.length > 0) {
+    lines.push("### Validation Warnings", "");
+    for (const issue of report.validationIssues) {
+      const category = issue.categoryId ? `${issue.categoryId}: ` : "";
+      lines.push(`- ${issue.code}: ${category}${issue.message}`);
+    }
+    lines.push("");
+  }
+
+  if (report.unmappedCoverageIds.length > 0) {
+    lines.push("### Unmapped Coverage IDs", "");
+    lines.push(report.unmappedCoverageIds.join(", "));
+    lines.push("");
+  }
+}
+
 export function renderQaCoverageMarkdownReport(inventory: QaCoverageInventory): string {
   const lines: string[] = [
     "# QA Coverage Inventory",
@@ -349,6 +413,8 @@ export function renderQaCoverageMarkdownReport(inventory: QaCoverageInventory):
     lines.push("");
   }
 
+  pushScorecardTaxonomyLines(lines, inventory.scorecardTaxonomy);
+
   if (inventory.overlappingCoverage.length > 0) {
     lines.push("## Overlap", "");
     pushFeatureLines(lines, inventory.overlappingCoverage);

extensions/qa-lab/src/docker-harness.test.ts

@@ -3,6 +3,7 @@ import { mkdtemp, readFile, rm } from "node:fs/promises";
 import os from "node:os";
 import path from "node:path";
 import { afterEach, describe, expect, it } from "vitest";
+import YAML from "yaml";
 import { buildQaDockerHarnessImage, writeQaDockerHarnessFiles } from "./docker-harness.js";
 
 const cleanups: Array<() => Promise<void>> = [];
@@ -13,6 +14,19 @@ afterEach(async () => {
   }
 });
 
+function parseComposeServices(compose: string) {
+  const parsed = YAML.parse(compose) as {
+    services?: Record<
+      string,
+      {
+        environment?: Record<string, string>;
+        volumes?: string[];
+      }
+    >;
+  };
+  return parsed.services ?? {};
+}
+
 describe("qa docker harness", () => {
   it("writes compose, env, config, and workspace scaffold files", async () => {
     const outputDir = await mkdtemp(path.join(os.tmpdir(), "qa-docker-test-"));
@@ -45,8 +59,21 @@ describe("qa docker harness", () => {
     }
 
     const compose = await readFile(path.join(outputDir, "docker-compose.qa.yml"), "utf8");
+    const services = parseComposeServices(compose);
     expect(compose).toContain("image: openclaw:qa-local-prebaked");
     expect(compose).toContain("qa-mock-openai:");
+    expect(services["qa-mock-openai"]?.environment).toMatchObject({
+      OPENCLAW_ENABLE_PRIVATE_QA_CLI: "1",
+      OPENCLAW_PROFILE: "",
+    });
+    expect(services["qa-mock-openai"]?.environment).not.toHaveProperty("OPENCLAW_CONFIG_PATH");
+    expect(services["qa-mock-openai"]?.volumes).toBeUndefined();
+    expect(services["qa-lab"]?.environment).toMatchObject({
+      OPENCLAW_ENABLE_PRIVATE_QA_CLI: "1",
+      OPENCLAW_CONFIG_PATH: "/opt/openclaw-scaffold/openclaw.json",
+      OPENCLAW_STATE_DIR: "/tmp/openclaw/state",
+    });
+    expect(services["qa-lab"]?.volumes).toContain("./state:/opt/openclaw-scaffold:ro");
     expect(compose).toContain('      - "127.0.0.1:18889:18789"');
     expect(compose).toContain('      - "127.0.0.1:43124:43123"');
     expect(compose).toContain(":/opt/openclaw-qa-lab-ui:ro");
@@ -75,13 +102,21 @@ describe("qa docker harness", () => {
     expect(envExample).toContain("QA_PROVIDER_BASE_URL=http://host.docker.internal:45123/v1");
     expect(envExample).toContain("QA_LAB_URL=http://127.0.0.1:43124");
 
-    const config = await readFile(path.join(outputDir, "state", "openclaw.json"), "utf8");
-    expect(config).toContain('"allowInsecureAuth": true');
-    expect(config).toContain('"pluginToolsMcpBridge": true');
-    expect(config).toContain('"openClawToolsMcpBridge": true');
-    expect(config).toContain("/app/dist/control-ui");
-    expect(config).toContain("C-3PO QA");
-    expect(config).toContain('"/tmp/openclaw/workspace"');
+    const configText = await readFile(path.join(outputDir, "state", "openclaw.json"), "utf8");
+    const config = JSON.parse(configText) as {
+      plugins?: {
+        allow?: string[];
+        entries?: Record<string, { enabled?: boolean }>;
+      };
+    };
+    expect(configText).toContain('"allowInsecureAuth": true');
+    expect(configText).toContain('"pluginToolsMcpBridge": true');
+    expect(configText).toContain('"openClawToolsMcpBridge": true');
+    expect(configText).toContain("/app/dist/control-ui");
+    expect(configText).toContain("C-3PO QA");
+    expect(configText).toContain('"/tmp/openclaw/workspace"');
+    expect(config.plugins?.allow).toContain("qa-lab");
+    expect(config.plugins?.entries?.["qa-lab"]?.enabled).toBe(true);
 
     const kickoff = await readFile(
       path.join(outputDir, "state", "seed-workspace", "QA_KICKOFF_TASK.md"),

extensions/qa-lab/src/docker-harness.ts

@@ -60,6 +60,9 @@ ${imageBlock}    pull_policy: never
       timeout: 5s
       retries: 6
       start_period: 3s
+    environment:
+      OPENCLAW_ENABLE_PRIVATE_QA_CLI: "1"
+      OPENCLAW_PROFILE: ""
     command:
       - node
       - dist/index.js
@@ -88,6 +91,9 @@ ${params.bindUiDist ? `      - ${qaLabUiMount}:${QA_LAB_UI_OVERLAY_DIR}:ro\n` :
       retries: 6
       start_period: 5s
     environment:
+      OPENCLAW_ENABLE_PRIVATE_QA_CLI: "1"
+      OPENCLAW_CONFIG_PATH: /opt/openclaw-scaffold/openclaw.json
+      OPENCLAW_STATE_DIR: /tmp/openclaw/state
       OPENCLAW_SKIP_GMAIL_WATCHER: "1"
       OPENCLAW_SKIP_BROWSER_CONTROL_SERVER: "1"
       OPENCLAW_SKIP_CANVAS_HOST: "1"

extensions/qa-lab/src/errors.ts

@@ -0,0 +1,34 @@
+// Qa Lab plugin module defines shared suite errors.
+export type QaSuiteArtifactErrorCode =
+  | "report_missing"
+  | "summary_read_failed"
+  | "summary_parse_failed"
+  | "summary_failure_count_missing"
+  | "summary_blocking_count_missing";
+
+export class QaSuiteArtifactError extends Error {
+  readonly code: QaSuiteArtifactErrorCode;
+
+  constructor(code: QaSuiteArtifactErrorCode, message: string, options?: { cause?: unknown }) {
+    super(message, options);
+    this.name = "QaSuiteArtifactError";
+    this.code = code;
+  }
+}
+
+export type QaSuiteInfraErrorCode =
+  | "agent_wait_failed"
+  | "gateway_startup_unhealthy"
+  | "gateway_ready_timeout"
+  | "qa_cli_timeout"
+  | "transport_ready_timeout";
+
+export class QaSuiteInfraError extends Error {
+  readonly code: QaSuiteInfraErrorCode;
+
+  constructor(code: QaSuiteInfraErrorCode, message: string, options?: { cause?: unknown }) {
+    super(message, options);
+    this.name = "QaSuiteInfraError";
+    this.code = code;
+  }
+}

extensions/qa-lab/src/evidence-summary.test.ts

@@ -0,0 +1,599 @@
+// Qa Lab tests cover QA evidence summary behavior.
+import { describe, expect, it } from "vitest";
+import {
+  QA_EVIDENCE_SUMMARY_KIND,
+  QA_EVIDENCE_FILENAME,
+  QA_EVIDENCE_SUMMARY_SCHEMA_VERSION,
+  buildLiveTransportEvidenceSummary,
+  buildPlaywrightEvidenceSummary,
+  buildQaSuiteEvidenceSummary,
+  buildVitestEvidenceSummary,
+  validateQaEvidenceSummaryJson,
+} from "./evidence-summary.js";
+
+describe("evidence summary", () => {
+  it("builds taxonomy-mapped QA suite evidence entries from catalog metadata", () => {
+    const evidence = buildQaSuiteEvidenceSummary({
+      artifactPaths: [
+        { kind: "summary", path: "qa-suite-summary.json" },
+        { kind: "report", path: "qa-suite-report.md" },
+      ],
+      scenarioDefinitions: [
+        {
+          id: "dm-chat-baseline",
+          title: "DM baseline conversation",
+          sourcePath: "qa/scenarios/channels/dm-chat-baseline.md",
+          surface: "dm",
+          coverage: {
+            primary: ["channels.dm"],
+            secondary: ["channels.qa-channel"],
+          },
+          runtimeParityTier: "standard",
+          docsRefs: ["docs/channels/qa-channel.md"],
+          codeRefs: ["extensions/qa-channel/src/gateway.ts"],
+        },
+      ],
+      channelId: "qa-channel",
+      env: {
+        OPENCLAW_QA_CHANNEL_DRIVER: "local-shim",
+        OPENCLAW_QA_REF: "abc123",
+      } as NodeJS.ProcessEnv,
+      generatedAt: "2026-06-07T12:00:00.000Z",
+      primaryModel: "mock-openai/gpt-5.5",
+      providerMode: "mock-openai",
+      scenarioResults: [{ name: "DM baseline conversation", status: "pass" }],
+    });
+
+    expect(validateQaEvidenceSummaryJson(evidence)).toEqual(evidence);
+    expect(evidence.kind).toBe(QA_EVIDENCE_SUMMARY_KIND);
+    expect(evidence.schemaVersion).toBe(QA_EVIDENCE_SUMMARY_SCHEMA_VERSION);
+    expect(evidence.entries).toHaveLength(1);
+    expect(evidence.entries[0]).toMatchObject({
+      test: {
+        kind: "qa-scenario",
+        id: "dm-chat-baseline",
+        title: "DM baseline conversation",
+        source: {
+          path: "qa/scenarios/channels/dm-chat-baseline.md",
+        },
+      },
+      mapping: {
+        profile: "smoke-ci",
+        coverage: [
+          {
+            id: "channels.dm",
+            role: "primary",
+            surfaceIds: ["dm"],
+            categoryIds: ["channels.dm"],
+          },
+          {
+            id: "channels.qa-channel",
+            role: "secondary",
+            surfaceIds: ["dm"],
+            categoryIds: [],
+          },
+        ],
+        refs: [
+          {
+            kind: "docs",
+            path: "docs/channels/qa-channel.md",
+          },
+          {
+            kind: "code",
+            path: "extensions/qa-channel/src/gateway.ts",
+          },
+        ],
+        runtimeParityTier: "standard",
+      },
+      execution: {
+        runner: "host",
+        provider: {
+          id: "openai",
+          live: false,
+          model: {
+            name: "gpt-5.5",
+            ref: "mock-openai/gpt-5.5",
+          },
+          fixture: "mock-openai",
+        },
+        channel: {
+          id: "qa-channel",
+          live: false,
+          driver: "local-shim",
+        },
+        packageSource: {
+          kind: "source-checkout",
+        },
+        environment: {
+          ref: "abc123",
+          os: process.platform,
+          nodeVersion: process.version,
+        },
+        artifacts: [
+          {
+            kind: "summary",
+            path: "qa-suite-summary.json",
+            source: "qa-suite",
+          },
+          {
+            kind: "report",
+            path: "qa-suite-report.md",
+            source: "qa-suite",
+          },
+        ],
+      },
+      result: {
+        status: "pass",
+      },
+    });
+  });
+
+  it("builds Telegram live transport evidence entries", () => {
+    const evidence = buildLiveTransportEvidenceSummary({
+      artifactPaths: [
+        { kind: "summary", path: QA_EVIDENCE_FILENAME },
+        { kind: "report", path: "telegram-qa-report.md" },
+        { kind: "transport-observations", path: "telegram-qa-observed-messages.json" },
+      ],
+      env: {
+        OPENCLAW_QA_RUNNER: "crabbox",
+      } as NodeJS.ProcessEnv,
+      generatedAt: "2026-06-07T12:05:00.000Z",
+      primaryModel: "openai/gpt-5.5",
+      providerMode: "live-frontier",
+      checks: [
+        {
+          id: "telegram-canary",
+          standardId: "canary",
+          title: "Telegram canary",
+          status: "fail",
+          details: "timed out waiting for SUT reply",
+          rttMs: 4321,
+        },
+      ],
+      transportId: "telegram",
+    });
+
+    expect(validateQaEvidenceSummaryJson(evidence)).toEqual(evidence);
+    expect(evidence.entries).toEqual([
+      expect.objectContaining({
+        test: {
+          kind: "live-transport-check",
+          id: "telegram-canary",
+          title: "Telegram canary",
+        },
+        mapping: {
+          profile: "release",
+          coverage: [
+            {
+              id: "channels.telegram.live",
+              role: "live-transport",
+              surfaceIds: ["channels.telegram"],
+              categoryIds: ["channels.telegram.live"],
+            },
+            {
+              id: "channels.telegram.canary",
+              role: "live-transport-standard",
+              surfaceIds: ["channels.telegram"],
+              categoryIds: ["channels.telegram.live"],
+            },
+          ],
+        },
+        execution: expect.objectContaining({
+          runner: "crabbox",
+          provider: {
+            id: "openai",
+            live: true,
+            model: {
+              name: "gpt-5.5",
+              ref: "openai/gpt-5.5",
+            },
+            auth: "live-frontier",
+          },
+          channel: {
+            id: "telegram",
+            live: true,
+            driver: "native",
+          },
+          artifacts: [
+            {
+              kind: "summary",
+              path: QA_EVIDENCE_FILENAME,
+              source: "telegram-live-transport",
+            },
+            {
+              kind: "report",
+              path: "telegram-qa-report.md",
+              source: "telegram-live-transport",
+            },
+            {
+              kind: "transport-observations",
+              path: "telegram-qa-observed-messages.json",
+              source: "telegram-live-transport",
+            },
+          ],
+        }),
+        result: {
+          status: "fail",
+          failure: {
+            reason: "timed out waiting for SUT reply",
+          },
+          timing: {
+            rttMs: 4321,
+          },
+        },
+      }),
+    ]);
+  });
+
+  it("builds Vitest runner evidence entries", () => {
+    const evidence = buildVitestEvidenceSummary({
+      artifactPaths: [
+        { kind: "runner-result", path: "vitest-results/runtime-boundary.vitest.json" },
+      ],
+      env: {
+        OPENCLAW_QA_REF: "abc123",
+      } as NodeJS.ProcessEnv,
+      generatedAt: "2026-06-07T12:06:00.000Z",
+      primaryModel: "mock-openai/gpt-5.5",
+      providerMode: "mock-openai",
+      targets: [
+        {
+          id: "runtime.agent-runner-boundary",
+          title: "Agent runner boundary integration tests",
+          sourcePath: "src/agents/agent-runner.e2e.test.ts",
+          coverageIds: ["runtime.agent-runner", "runtime.delivery"],
+          surfaceIds: ["agent-runtime-and-provider-execution"],
+          categoryIds: ["agent-runtime-and-provider-execution.agent-turn-execution"],
+          codeRefs: ["src/agents/agent-runner.ts"],
+        },
+      ],
+      results: [
+        {
+          id: "runtime.agent-runner-boundary",
+          status: "pass",
+          durationMs: 1234,
+        },
+      ],
+    });
+
+    expect(validateQaEvidenceSummaryJson(evidence)).toEqual(evidence);
+    expect(evidence.entries).toEqual([
+      expect.objectContaining({
+        test: {
+          kind: "vitest-test",
+          id: "runtime.agent-runner-boundary",
+          title: "Agent runner boundary integration tests",
+          source: {
+            path: "src/agents/agent-runner.e2e.test.ts",
+          },
+        },
+        mapping: {
+          profile: "smoke-ci",
+          coverage: [
+            {
+              id: "runtime.agent-runner",
+              role: "primary",
+              surfaceIds: ["agent-runtime-and-provider-execution"],
+              categoryIds: ["agent-runtime-and-provider-execution.agent-turn-execution"],
+            },
+            {
+              id: "runtime.delivery",
+              role: "primary",
+              surfaceIds: ["agent-runtime-and-provider-execution"],
+              categoryIds: ["agent-runtime-and-provider-execution.agent-turn-execution"],
+            },
+          ],
+          refs: [
+            {
+              kind: "code",
+              path: "src/agents/agent-runner.ts",
+            },
+          ],
+        },
+        execution: expect.objectContaining({
+          runner: "vitest",
+          provider: expect.objectContaining({
+            live: false,
+            fixture: "mock-openai",
+          }),
+          artifacts: [
+            {
+              kind: "runner-result",
+              path: "vitest-results/runtime-boundary.vitest.json",
+              source: "vitest",
+            },
+          ],
+        }),
+        result: {
+          status: "pass",
+          timing: {
+            wallMs: 1234,
+          },
+        },
+      }),
+    ]);
+  });
+
+  it("builds Playwright runner evidence entries", () => {
+    const evidence = buildPlaywrightEvidenceSummary({
+      artifactPaths: [
+        { kind: "runner-result", path: "playwright-results/control-ui.json" },
+        { kind: "report", path: "playwright-report/index.html" },
+      ],
+      env: {
+        GITHUB_SHA: "def456",
+      } as NodeJS.ProcessEnv,
+      generatedAt: "2026-06-07T12:07:00.000Z",
+      primaryModel: "mock-openai/gpt-5.5",
+      providerMode: "mock-openai",
+      targets: [
+        {
+          id: "control-ui.browser-run",
+          title: "Control UI browser workflow",
+          sourcePath: "ui/control-ui.e2e.test.ts",
+          coverageIds: ["control-ui.browser"],
+          surfaceIds: ["browser-control-ui-and-webchat"],
+          categoryIds: ["browser-control-ui-and-webchat.browser-ui"],
+          docsRefs: ["docs/concepts/qa-e2e-automation.md"],
+          codeRefs: ["ui/"],
+        },
+      ],
+      results: [
+        {
+          id: "control-ui.browser-run",
+          status: "fail",
+          durationMs: 2300,
+          failureMessage: "locator timed out",
+        },
+      ],
+    });
+
+    expect(validateQaEvidenceSummaryJson(evidence)).toEqual(evidence);
+    expect(evidence.entries[0]).toMatchObject({
+      test: {
+        kind: "playwright-test",
+        id: "control-ui.browser-run",
+        title: "Control UI browser workflow",
+        source: {
+          path: "ui/control-ui.e2e.test.ts",
+        },
+      },
+      mapping: {
+        coverage: [
+          {
+            id: "control-ui.browser",
+            role: "primary",
+            surfaceIds: ["browser-control-ui-and-webchat"],
+            categoryIds: ["browser-control-ui-and-webchat.browser-ui"],
+          },
+        ],
+        refs: [
+          {
+            kind: "docs",
+            path: "docs/concepts/qa-e2e-automation.md",
+          },
+          {
+            kind: "code",
+            path: "ui/",
+          },
+        ],
+      },
+      execution: {
+        runner: "playwright",
+        artifacts: [
+          {
+            kind: "runner-result",
+            path: "playwright-results/control-ui.json",
+            source: "playwright",
+          },
+          {
+            kind: "report",
+            path: "playwright-report/index.html",
+            source: "playwright",
+          },
+        ],
+      },
+      result: {
+        status: "fail",
+        failure: {
+          reason: "locator timed out",
+        },
+        timing: {
+          wallMs: 2300,
+        },
+      },
+    });
+  });
+
+  it("carries profile env values without hardcoding taxonomy mapping ids", () => {
+    const evidence = buildQaSuiteEvidenceSummary({
+      artifactPaths: [{ kind: "summary", path: "qa-suite-summary.json" }],
+      scenarioDefinitions: [
+        {
+          id: "dm-chat-baseline",
+          title: "DM baseline conversation",
+          surface: "dm",
+          coverage: {
+            primary: ["channels.dm"],
+          },
+        },
+      ],
+      channelId: "qa-channel",
+      env: {
+        OPENCLAW_QA_PROFILE: "experimental-profile",
+      } as NodeJS.ProcessEnv,
+      generatedAt: "2026-06-07T12:09:00.000Z",
+      primaryModel: "mock-openai/gpt-5.5",
+      providerMode: "mock-openai",
+      scenarioResults: [{ name: "DM baseline conversation", status: "pass" }],
+    });
+
+    expect(evidence.entries[0]?.mapping.profile).toBe("experimental-profile");
+  });
+
+  it("keeps mock non-OpenAI model refs attributed to their model provider", () => {
+    const evidence = buildQaSuiteEvidenceSummary({
+      artifactPaths: [{ kind: "summary", path: "qa-suite-summary.json" }],
+      scenarioDefinitions: [
+        {
+          id: "anthropic-parity",
+          title: "Anthropic parity",
+          surface: "runtime",
+          coverage: {
+            primary: ["providers.anthropic"],
+          },
+        },
+      ],
+      channelId: "qa-channel",
+      generatedAt: "2026-06-07T12:10:00.000Z",
+      primaryModel: "anthropic/claude-opus-4-8",
+      providerMode: "mock-openai",
+      scenarioResults: [{ name: "Anthropic parity", status: "pass" }],
+    });
+
+    expect(evidence.entries[0]?.execution.provider).toMatchObject({
+      id: "anthropic",
+      model: {
+        name: "claude-opus-4-8",
+        ref: "anthropic/claude-opus-4-8",
+      },
+    });
+    expect(evidence.entries[0]).toMatchObject({
+      execution: {
+        provider: {
+          live: false,
+          fixture: "mock-openai",
+        },
+      },
+    });
+  });
+
+  it("uses explicit package provenance from package runners", () => {
+    const evidence = buildLiveTransportEvidenceSummary({
+      artifactPaths: [{ kind: "summary", path: QA_EVIDENCE_FILENAME }],
+      generatedAt: "2026-06-07T12:15:00.000Z",
+      packageSource: {
+        kind: "packed-tarball",
+        spec: "/tmp/openclaw.tgz",
+        sha: "abc123",
+      },
+      primaryModel: "openai/gpt-5.5",
+      providerMode: "live-frontier",
+      checks: [
+        {
+          id: "telegram-canary",
+          title: "Telegram canary",
+          details: "Canary passed.",
+          standardId: "canary",
+          status: "pass",
+        },
+      ],
+      transportId: "telegram",
+    });
+
+    expect(evidence.entries[0]?.execution.packageSource).toEqual({
+      kind: "packed-tarball",
+      spec: "/tmp/openclaw.tgz",
+      sha: "abc123",
+    });
+  });
+
+  it("derives package provenance from generic QA evidence env", () => {
+    const evidence = buildLiveTransportEvidenceSummary({
+      artifactPaths: [{ kind: "summary", path: QA_EVIDENCE_FILENAME }],
+      env: {
+        OPENCLAW_QA_PACKAGE_SOURCE: "openclaw@beta",
+        OPENCLAW_QA_PACKAGE_SOURCE_KIND: "npm-package",
+        OPENCLAW_QA_PACKAGE_SOURCE_SHA: "def456",
+      } as NodeJS.ProcessEnv,
+      generatedAt: "2026-06-07T12:15:00.000Z",
+      primaryModel: "openai/gpt-5.5",
+      providerMode: "live-frontier",
+      checks: [
+        {
+          id: "telegram-canary",
+          title: "Telegram canary",
+          details: "Canary passed.",
+          standardId: "canary",
+          status: "pass",
+        },
+      ],
+      transportId: "telegram",
+    });
+
+    expect(evidence.entries[0]?.execution.packageSource).toEqual({
+      kind: "npm-package",
+      spec: "openclaw@beta",
+      sha: "def456",
+    });
+  });
+
+  it("does not infer package provenance from runner-specific env", () => {
+    const evidence = buildLiveTransportEvidenceSummary({
+      artifactPaths: [{ kind: "summary", path: QA_EVIDENCE_FILENAME }],
+      env: {
+        OPENCLAW_NPM_TELEGRAM_INSTALL_SOURCE: "openclaw@beta",
+      } as NodeJS.ProcessEnv,
+      generatedAt: "2026-06-07T12:16:00.000Z",
+      primaryModel: "openai/gpt-5.5",
+      providerMode: "live-frontier",
+      checks: [
+        {
+          id: "telegram-canary",
+          title: "Telegram canary",
+          details: "Canary passed.",
+          standardId: "canary",
+          status: "pass",
+        },
+      ],
+      transportId: "telegram",
+    });
+
+    expect(evidence.entries[0]?.execution.packageSource).toEqual({
+      kind: "source-checkout",
+      spec: undefined,
+      sha: undefined,
+    });
+  });
+
+  it("keeps live transport check artifacts on the owning entry", () => {
+    const evidence = buildLiveTransportEvidenceSummary({
+      artifactPaths: [
+        { kind: "summary", path: QA_EVIDENCE_FILENAME },
+        { kind: "report", path: "discord-qa-report.md" },
+      ],
+      generatedAt: "2026-06-07T12:20:00.000Z",
+      primaryModel: "openai/gpt-5.5",
+      providerMode: "live-frontier",
+      checks: [
+        {
+          artifactPaths: {
+            screenshot: ".artifacts/discord/status.png",
+            video: ".artifacts/discord/status.mp4",
+          },
+          id: "discord-status-reactions-tool-only",
+          title: "Discord status reactions",
+          details: "Status reaction observed.",
+          status: "pass",
+        },
+      ],
+      transportId: "discord",
+    });
+
+    expect(evidence.entries[0]?.execution.artifacts).toEqual(
+      expect.arrayContaining([
+        {
+          kind: "screenshot",
+          path: ".artifacts/discord/status.png",
+          source: "discord-live-transport:discord-status-reactions-tool-only",
+        },
+        {
+          kind: "video",
+          path: ".artifacts/discord/status.mp4",
+          source: "discord-live-transport:discord-status-reactions-tool-only",
+        },
+      ]),
+    );
+  });
+});

extensions/qa-lab/src/evidence-summary.ts

@@ -0,0 +1,701 @@
+// Qa Lab plugin module implements QA evidence summary behavior.
+import { z } from "zod";
+import { splitQaModelRef } from "./model-selection.js";
+import { getQaProvider, type QaProviderMode } from "./providers/index.js";
+
+export const QA_EVIDENCE_SUMMARY_KIND = "openclaw.qa.evidence-summary";
+export const QA_EVIDENCE_FILENAME = "qa-evidence.json";
+export const QA_EVIDENCE_SUMMARY_SCHEMA_VERSION = 2;
+
+const qaEvidenceStatusSchema = z.enum(["pass", "fail", "blocked", "skipped"]);
+const nonEmptyStringSchema = z.string().trim().min(1);
+const nullableStringSchema = nonEmptyStringSchema.nullable();
+const qaEvidenceProfileIdSchema = nonEmptyStringSchema;
+const qaEvidenceIdSchema = z.object({ id: nonEmptyStringSchema });
+
+const qaEvidenceProviderSchema = z
+  .object({
+    id: nonEmptyStringSchema,
+    live: z.boolean(),
+    model: z
+      .object({
+        name: nullableStringSchema,
+        ref: nullableStringSchema,
+      })
+      .strict(),
+    fixture: nonEmptyStringSchema.optional(),
+    auth: nonEmptyStringSchema.optional(),
+  })
+  .strict();
+
+const qaEvidenceChannelSchema = z
+  .object({
+    id: nonEmptyStringSchema,
+    live: z.boolean(),
+    driver: nonEmptyStringSchema.optional(),
+  })
+  .strict();
+
+const qaEvidenceEnvironmentSchema = z
+  .object({
+    ref: nullableStringSchema,
+    os: nonEmptyStringSchema,
+    nodeVersion: nonEmptyStringSchema,
+  })
+  .strict();
+
+const qaEvidencePackageSourceSchema = z
+  .object({
+    kind: nonEmptyStringSchema,
+    spec: nonEmptyStringSchema.optional(),
+    sha: nonEmptyStringSchema.optional(),
+  })
+  .strict();
+
+const qaEvidenceFailureSchema = z
+  .object({
+    class: nonEmptyStringSchema.optional(),
+    reason: nonEmptyStringSchema,
+  })
+  .strict();
+
+const qaEvidenceTimingSchema = z
+  .object({
+    wallMs: z.number().finite().positive().optional(),
+    rttMs: z.number().finite().positive().optional(),
+    avgMs: z.number().finite().positive().optional(),
+    p50Ms: z.number().finite().positive().optional(),
+    p95Ms: z.number().finite().positive().optional(),
+    maxMs: z.number().finite().positive().optional(),
+    samples: z.number().int().positive().optional(),
+    failedSamples: z.number().int().nonnegative().optional(),
+  })
+  .strict();
+
+const qaEvidenceTestSchema = z
+  .object({
+    kind: nonEmptyStringSchema,
+    id: nonEmptyStringSchema,
+    title: nonEmptyStringSchema,
+    source: z
+      .object({
+        path: nonEmptyStringSchema,
+      })
+      .strict()
+      .optional(),
+  })
+  .strict();
+
+const qaEvidenceRefSchema = z
+  .object({
+    kind: nonEmptyStringSchema,
+    path: nonEmptyStringSchema,
+  })
+  .strict();
+
+const qaEvidenceCoverageSchema = qaEvidenceIdSchema
+  .extend({
+    role: nonEmptyStringSchema,
+    surfaceIds: z.array(nonEmptyStringSchema),
+    categoryIds: z.array(nonEmptyStringSchema),
+  })
+  .strict();
+
+const qaEvidenceMappingSchema = z
+  .object({
+    profile: qaEvidenceProfileIdSchema,
+    coverage: z.array(qaEvidenceCoverageSchema),
+    refs: z.array(qaEvidenceRefSchema).optional(),
+    runtimeParityTier: nonEmptyStringSchema.optional(),
+  })
+  .strict();
+
+const qaEvidenceArtifactSchema = z
+  .object({
+    kind: nonEmptyStringSchema,
+    path: nonEmptyStringSchema,
+    source: nonEmptyStringSchema,
+  })
+  .strict();
+
+const qaEvidenceExecutionSchema = z
+  .object({
+    runner: nonEmptyStringSchema,
+    environment: qaEvidenceEnvironmentSchema,
+    provider: qaEvidenceProviderSchema,
+    channel: qaEvidenceChannelSchema.optional(),
+    packageSource: qaEvidencePackageSourceSchema,
+    artifacts: z.array(qaEvidenceArtifactSchema),
+  })
+  .strict();
+
+const qaEvidenceResultSchema = z
+  .object({
+    status: qaEvidenceStatusSchema,
+    failure: qaEvidenceFailureSchema.optional(),
+    timing: qaEvidenceTimingSchema.optional(),
+  })
+  .strict();
+
+export const qaEvidenceSummaryEntrySchema = z
+  .object({
+    test: qaEvidenceTestSchema,
+    mapping: qaEvidenceMappingSchema,
+    execution: qaEvidenceExecutionSchema,
+    result: qaEvidenceResultSchema,
+  })
+  .strict();
+
+export const qaEvidenceSummarySchema = z
+  .object({
+    kind: z.literal(QA_EVIDENCE_SUMMARY_KIND),
+    schemaVersion: z.literal(QA_EVIDENCE_SUMMARY_SCHEMA_VERSION),
+    generatedAt: nonEmptyStringSchema,
+    entries: z.array(qaEvidenceSummaryEntrySchema),
+  })
+  .strict();
+
+export type QaEvidenceProfile = z.infer<typeof qaEvidenceProfileIdSchema>;
+export type QaEvidenceStatus = z.infer<typeof qaEvidenceStatusSchema>;
+export type QaEvidenceTiming = z.infer<typeof qaEvidenceTimingSchema>;
+export type QaEvidencePackageSource = z.infer<typeof qaEvidencePackageSourceSchema>;
+export type QaEvidenceSummaryEntry = z.infer<typeof qaEvidenceSummaryEntrySchema>;
+export type QaEvidenceSummaryJson = z.infer<typeof qaEvidenceSummarySchema>;
+
+type QaEvidenceStatusInput = QaEvidenceStatus | "skip";
+
+type QaEvidenceScenarioDefinitionInput = {
+  id: string;
+  title: string;
+  sourcePath?: string;
+  surface?: string;
+  surfaces?: readonly string[];
+  category?: string;
+  coverage?: {
+    primary?: readonly string[];
+    secondary?: readonly string[];
+  };
+  runtimeParityTier?: string;
+  docsRefs?: readonly string[];
+  codeRefs?: readonly string[];
+};
+
+type QaEvidenceScenarioResultInput = {
+  name: string;
+  status: QaEvidenceStatusInput;
+  details?: string;
+  rttMs?: number;
+  rttMeasurement?: {
+    finalMatchedReplyRttMs?: number;
+  };
+};
+
+type QaEvidenceLiveTransportCheckInput = {
+  id: string;
+  title: string;
+  status: QaEvidenceStatusInput;
+  details: string;
+  rttMs?: number;
+  rttMeasurement?: {
+    finalMatchedReplyRttMs?: number;
+  };
+  // Here "standard" means a taxonomy-backed requirement standard, not the default lane.
+  standardId?: string;
+  artifactPaths?: Readonly<Record<string, string>>;
+};
+
+type QaEvidenceRttInput = Pick<QaEvidenceScenarioResultInput, "rttMeasurement" | "rttMs">;
+
+type QaEvidenceTestTargetInput = {
+  id: string;
+  title: string;
+  sourcePath: string;
+  coverageIds: readonly string[];
+  surfaceIds: readonly string[];
+  categoryIds: readonly string[];
+  docsRefs?: readonly string[];
+  codeRefs?: readonly string[];
+};
+
+type QaEvidenceTestResultInput = {
+  id?: string;
+  title?: string;
+  sourcePath?: string;
+  status: QaEvidenceStatusInput;
+  durationMs?: number;
+  failureMessage?: string;
+};
+
+type QaEvidenceArtifactInput = {
+  kind: string;
+  path: string;
+};
+
+type QaEvidenceBuildBase = {
+  artifactPaths: readonly QaEvidenceArtifactInput[];
+  env?: NodeJS.ProcessEnv;
+  generatedAt: string;
+  primaryModel: string;
+  providerMode: QaProviderMode;
+  channelDriver?: string;
+  packageSource?: QaEvidencePackageSource;
+  profile?: QaEvidenceProfile;
+  runner?: string;
+};
+
+function buildQaEvidenceRefs(params: {
+  docsRefs?: readonly string[];
+  codeRefs?: readonly string[];
+}) {
+  const buildRef = (kind: "docs" | "code", refPath: string) => {
+    const ref = {
+      kind,
+      path: refPath,
+    };
+    return ref;
+  };
+  const refs = [
+    ...(params.docsRefs ?? []).map((path) => buildRef("docs", path)),
+    ...(params.codeRefs ?? []).map((path) => buildRef("code", path)),
+  ];
+  return [...new Map(refs.map((ref) => [`${ref.kind}:${ref.path}`, ref])).values()];
+}
+
+function buildQaEvidenceCoverage(params: {
+  primaryIds?: readonly string[];
+  secondaryIds?: readonly string[];
+  surfaceIds?: readonly string[];
+  categoryIds?: readonly string[];
+}) {
+  const surfaceIds = uniqueSortedStrings(params.surfaceIds ?? []);
+  const categoryIds = uniqueSortedStrings(params.categoryIds ?? []);
+  const buildCoverage = (id: string, role: "primary" | "secondary") => ({
+    id,
+    role,
+    surfaceIds,
+    categoryIds: role === "primary" ? categoryIds : [],
+  });
+  return [
+    ...uniqueSortedStrings(params.primaryIds ?? []).map((id) => buildCoverage(id, "primary")),
+    ...uniqueSortedStrings(params.secondaryIds ?? []).map((id) => buildCoverage(id, "secondary")),
+  ];
+}
+
+function buildQaEvidenceArtifacts(paths: readonly QaEvidenceArtifactInput[], source: string) {
+  return paths.map((artifact) => ({
+    kind: artifact.kind,
+    path: artifact.path,
+    source,
+  }));
+}
+
+function buildQaEvidenceNamedArtifacts(paths: Readonly<Record<string, string>>, source: string) {
+  return Object.entries(paths).map(([kind, artifactPath]) => ({
+    kind,
+    path: artifactPath,
+    source,
+  }));
+}
+
+function uniqueSortedStrings(values: readonly (string | undefined)[]) {
+  return [...new Set(values.map((value) => value?.trim()).filter(Boolean) as string[])].toSorted(
+    (left, right) => left.localeCompare(right),
+  );
+}
+
+function resolveQaEvidenceProfile(params: {
+  env?: NodeJS.ProcessEnv;
+  fallback: QaEvidenceProfile;
+  explicit?: QaEvidenceProfile;
+}) {
+  if (params.explicit) {
+    const explicit = params.explicit.trim();
+    if (!explicit) {
+      throw new Error("evidence profile must be a non-empty string.");
+    }
+    return explicit;
+  }
+
+  const envProfiles = [
+    ["OPENCLAW_E2E_PROFILE", params.env?.OPENCLAW_E2E_PROFILE],
+    ["OPENCLAW_QA_PROFILE", params.env?.OPENCLAW_QA_PROFILE],
+  ] as const;
+  for (const [, value] of envProfiles) {
+    const normalized = value?.trim();
+    if (!normalized) {
+      continue;
+    }
+    return normalized;
+  }
+
+  return params.fallback;
+}
+
+function resolveQaEvidenceRunner(params: { env?: NodeJS.ProcessEnv; fallback?: string }) {
+  return params.env?.OPENCLAW_QA_RUNNER?.trim() || params.fallback || "host";
+}
+
+function resolveQaEvidenceChannelDriver(params: { env?: NodeJS.ProcessEnv; fallback?: string }) {
+  const id =
+    params.fallback?.trim() ||
+    params.env?.OPENCLAW_QA_CHANNEL_DRIVER?.trim() ||
+    params.env?.OPENCLAW_E2E_CHANNEL_DRIVER?.trim();
+  return id ? { id } : undefined;
+}
+
+function resolveQaEvidenceEnvironment(env: NodeJS.ProcessEnv | undefined) {
+  return {
+    ref: env?.OPENCLAW_QA_REF?.trim() || env?.GITHUB_SHA?.trim() || null,
+    os: process.platform,
+    nodeVersion: process.version,
+  };
+}
+
+function resolveQaEvidencePackageSource(env: NodeJS.ProcessEnv | undefined) {
+  const spec = env?.OPENCLAW_QA_PACKAGE_SOURCE?.trim() || undefined;
+  const sha = env?.OPENCLAW_QA_PACKAGE_SOURCE_SHA?.trim() || undefined;
+  const explicitKind = env?.OPENCLAW_QA_PACKAGE_SOURCE_KIND?.trim();
+  const kind =
+    explicitKind ||
+    (spec && spec.endsWith(".tgz") ? "packed-tarball" : spec ? "npm-package" : "source-checkout");
+  return {
+    kind,
+    spec,
+    sha,
+  };
+}
+
+function resolveQaEvidenceBuildPackageSource(params: QaEvidenceBuildBase) {
+  return params.packageSource ?? resolveQaEvidencePackageSource(params.env);
+}
+
+function buildQaEvidenceProvider(params: { providerMode: QaProviderMode; primaryModel: string }) {
+  const provider = getQaProvider(params.providerMode);
+  const split = splitQaModelRef(params.primaryModel);
+  const providerShape = {
+    id: split?.provider ?? params.providerMode,
+    model: {
+      name: split?.model ?? null,
+      ref: params.primaryModel || null,
+    },
+  };
+  if (provider.kind === "live") {
+    return {
+      ...providerShape,
+      live: true,
+      auth: params.providerMode,
+    };
+  }
+  const mockProviderId =
+    split?.provider && split.provider !== params.providerMode
+      ? split.provider
+      : params.providerMode === "mock-openai"
+        ? "openai"
+        : (split?.provider ?? params.providerMode);
+  return {
+    ...providerShape,
+    id: mockProviderId,
+    live: false,
+    fixture: params.providerMode,
+  };
+}
+
+function normalizeQaEvidenceStatus(status: QaEvidenceStatusInput): QaEvidenceStatus {
+  return status === "skip" ? "skipped" : status;
+}
+
+function failureForResult(result: {
+  details?: string;
+  failureMessage?: string;
+  status: QaEvidenceStatusInput;
+}) {
+  const status = normalizeQaEvidenceStatus(result.status);
+  if (status === "pass") {
+    return undefined;
+  }
+  return {
+    reason: result.details?.trim() || result.failureMessage?.trim() || `${status} test`,
+  };
+}
+
+function timingForRttResult(check: QaEvidenceRttInput) {
+  const rttMs = check.rttMeasurement?.finalMatchedReplyRttMs ?? check.rttMs;
+  return typeof rttMs === "number" && Number.isFinite(rttMs) && rttMs > 0 ? { rttMs } : undefined;
+}
+
+function timingForTestResult(result: QaEvidenceTestResultInput) {
+  return typeof result.durationMs === "number" &&
+    Number.isFinite(result.durationMs) &&
+    result.durationMs > 0
+    ? { wallMs: result.durationMs }
+    : undefined;
+}
+
+function resultForEvidence(
+  result: { details?: string; failureMessage?: string; status: QaEvidenceStatusInput },
+  timing?: QaEvidenceTiming,
+) {
+  return {
+    status: normalizeQaEvidenceStatus(result.status),
+    failure: failureForResult(result),
+    timing,
+  };
+}
+
+function buildQaEvidenceSummary(params: {
+  entries: QaEvidenceSummaryEntry[];
+  generatedAt: string;
+}): QaEvidenceSummaryJson {
+  return qaEvidenceSummarySchema.parse({
+    kind: QA_EVIDENCE_SUMMARY_KIND,
+    schemaVersion: QA_EVIDENCE_SUMMARY_SCHEMA_VERSION,
+    generatedAt: params.generatedAt,
+    entries: params.entries,
+  });
+}
+
+export function validateQaEvidenceSummaryJson(summary: unknown): QaEvidenceSummaryJson {
+  return qaEvidenceSummarySchema.parse(summary);
+}
+
+export function buildQaSuiteEvidenceSummary(
+  params: QaEvidenceBuildBase & {
+    channelId: string;
+    scenarioDefinitions: readonly QaEvidenceScenarioDefinitionInput[];
+    scenarioResults: readonly QaEvidenceScenarioResultInput[];
+  },
+): QaEvidenceSummaryJson {
+  const provider = buildQaEvidenceProvider(params);
+  const environment = resolveQaEvidenceEnvironment(params.env);
+  const packageSource = resolveQaEvidenceBuildPackageSource(params);
+  const runner = resolveQaEvidenceRunner({ env: params.env, fallback: params.runner });
+  const profile = resolveQaEvidenceProfile({
+    env: params.env,
+    fallback: provider.live ? "release" : "smoke-ci",
+    explicit: params.profile,
+  });
+  const channelDriver = resolveQaEvidenceChannelDriver({
+    env: params.env,
+    fallback: params.channelDriver,
+  });
+  const entries = params.scenarioResults.map((result, index): QaEvidenceSummaryEntry => {
+    const scenario = params.scenarioDefinitions[index];
+    const primaryCoverageIds = uniqueSortedStrings(scenario?.coverage?.primary ?? []);
+    const coverageIds = uniqueSortedStrings([
+      ...(scenario?.coverage?.primary ?? []),
+      ...(scenario?.coverage?.secondary ?? []),
+    ]);
+    const surfaceIds = uniqueSortedStrings(
+      scenario?.surfaces && scenario.surfaces.length > 0 ? scenario.surfaces : [scenario?.surface],
+    );
+    const runtimeParityTier = scenario?.runtimeParityTier;
+    const testId = scenario?.id ?? `scenario-${index + 1}`;
+    const refs = buildQaEvidenceRefs({
+      docsRefs: scenario?.docsRefs,
+      codeRefs: scenario?.codeRefs,
+    });
+    const timing = timingForRttResult(result);
+    return {
+      test: {
+        kind: "qa-scenario",
+        id: testId,
+        title: scenario?.title ?? result.name,
+        source: scenario?.sourcePath ? { path: scenario.sourcePath } : undefined,
+      },
+      mapping: {
+        profile,
+        coverage: buildQaEvidenceCoverage({
+          primaryIds: primaryCoverageIds,
+          secondaryIds: coverageIds.filter(
+            (coverageId) => !primaryCoverageIds.includes(coverageId),
+          ),
+          surfaceIds,
+          categoryIds: uniqueSortedStrings([scenario?.category, ...primaryCoverageIds]),
+        }),
+        refs: refs.length > 0 ? refs : undefined,
+        runtimeParityTier,
+      },
+      execution: {
+        runner,
+        environment,
+        provider,
+        channel: {
+          id: params.channelId,
+          live: false,
+          driver: channelDriver?.id,
+        },
+        packageSource,
+        artifacts: buildQaEvidenceArtifacts(params.artifactPaths, "qa-suite"),
+      },
+      result: resultForEvidence(result, timing),
+    };
+  });
+  return buildQaEvidenceSummary({ generatedAt: params.generatedAt, entries });
+}
+
+function buildTestRunnerEvidenceSummary(
+  params: QaEvidenceBuildBase & {
+    defaultRunner: string;
+    testKind: string;
+    targets: readonly QaEvidenceTestTargetInput[];
+    results: readonly QaEvidenceTestResultInput[];
+  },
+): QaEvidenceSummaryJson {
+  const provider = buildQaEvidenceProvider(params);
+  const environment = resolveQaEvidenceEnvironment(params.env);
+  const packageSource = resolveQaEvidenceBuildPackageSource(params);
+  const runner = resolveQaEvidenceRunner({
+    env: params.env,
+    fallback: params.runner ?? params.defaultRunner,
+  });
+  const profile = resolveQaEvidenceProfile({
+    env: params.env,
+    fallback: provider.live ? "release" : "smoke-ci",
+    explicit: params.profile,
+  });
+  const targetById = new Map(params.targets.map((target) => [target.id, target]));
+  const targetByPath = new Map(params.targets.map((target) => [target.sourcePath, target]));
+  const entries = params.results.map((result, index): QaEvidenceSummaryEntry => {
+    const target = result.id
+      ? targetById.get(result.id)
+      : result.sourcePath
+        ? targetByPath.get(result.sourcePath)
+        : undefined;
+    const fallbackId = result.id ?? result.sourcePath ?? `test-${index + 1}`;
+    const sourcePath = target?.sourcePath ?? result.sourcePath;
+    const refs = buildQaEvidenceRefs({
+      docsRefs: target?.docsRefs,
+      codeRefs: target?.codeRefs,
+    });
+    const timing = timingForTestResult(result);
+    return {
+      test: {
+        kind: params.testKind,
+        id: target?.id ?? fallbackId,
+        title: target?.title ?? result.title ?? fallbackId,
+        source: sourcePath ? { path: sourcePath } : undefined,
+      },
+      mapping: {
+        profile,
+        coverage: buildQaEvidenceCoverage({
+          primaryIds: target?.coverageIds ?? [],
+          surfaceIds: target?.surfaceIds ?? [],
+          categoryIds: target?.categoryIds ?? [],
+        }),
+        refs: refs.length > 0 ? refs : undefined,
+      },
+      execution: {
+        runner,
+        environment,
+        provider,
+        packageSource,
+        artifacts: buildQaEvidenceArtifacts(params.artifactPaths, runner),
+      },
+      result: resultForEvidence(result, timing),
+    };
+  });
+  return buildQaEvidenceSummary({ generatedAt: params.generatedAt, entries });
+}
+
+export function buildVitestEvidenceSummary(
+  params: QaEvidenceBuildBase & {
+    targets: readonly QaEvidenceTestTargetInput[];
+    results: readonly QaEvidenceTestResultInput[];
+  },
+): QaEvidenceSummaryJson {
+  return buildTestRunnerEvidenceSummary({
+    ...params,
+    defaultRunner: "vitest",
+    testKind: "vitest-test",
+    runner: params.runner ?? "vitest",
+  });
+}
+
+export function buildPlaywrightEvidenceSummary(
+  params: QaEvidenceBuildBase & {
+    targets: readonly QaEvidenceTestTargetInput[];
+    results: readonly QaEvidenceTestResultInput[];
+  },
+): QaEvidenceSummaryJson {
+  return buildTestRunnerEvidenceSummary({
+    ...params,
+    defaultRunner: "playwright",
+    testKind: "playwright-test",
+    runner: params.runner ?? "playwright",
+  });
+}
+
+export function buildLiveTransportEvidenceSummary(
+  params: QaEvidenceBuildBase & {
+    checks: readonly QaEvidenceLiveTransportCheckInput[];
+    transportId: string;
+  },
+): QaEvidenceSummaryJson {
+  const provider = buildQaEvidenceProvider(params);
+  const environment = resolveQaEvidenceEnvironment(params.env);
+  const packageSource = resolveQaEvidenceBuildPackageSource(params);
+  const runner = resolveQaEvidenceRunner({ env: params.env, fallback: params.runner });
+  const profile = resolveQaEvidenceProfile({
+    env: params.env,
+    fallback: "release",
+    explicit: params.profile,
+  });
+  const channelDriver = resolveQaEvidenceChannelDriver({
+    env: params.env,
+    fallback: params.channelDriver ?? "native",
+  }) ?? { id: "native" };
+  const entries = params.checks.map((check): QaEvidenceSummaryEntry => {
+    const testId = check.id;
+    const standardCoverageId = check.standardId
+      ? `channels.${params.transportId}.${check.standardId}`
+      : undefined;
+    const coverage = [
+      {
+        id: `channels.${params.transportId}.live`,
+        role: "live-transport",
+        surfaceIds: [`channels.${params.transportId}`],
+        categoryIds: [`channels.${params.transportId}.live`],
+      },
+    ];
+    if (standardCoverageId) {
+      coverage.push({
+        id: standardCoverageId,
+        role: "live-transport-standard",
+        surfaceIds: [`channels.${params.transportId}`],
+        categoryIds: [`channels.${params.transportId}.live`],
+      });
+    }
+    const timing = timingForRttResult(check);
+    return {
+      test: {
+        kind: "live-transport-check",
+        id: testId,
+        title: check.title,
+      },
+      mapping: {
+        profile,
+        coverage,
+      },
+      execution: {
+        runner,
+        environment,
+        provider,
+        channel: {
+          id: params.transportId,
+          live: true,
+          driver: channelDriver.id,
+        },
+        packageSource,
+        artifacts: [
+          ...buildQaEvidenceArtifacts(params.artifactPaths, `${params.transportId}-live-transport`),
+          ...buildQaEvidenceNamedArtifacts(
+            check.artifactPaths ?? {},
+            `${params.transportId}-live-transport:${testId}`,
+          ),
+        ],
+      },
+      result: resultForEvidence(check, timing),
+    };
+  });
+  return buildQaEvidenceSummary({ generatedAt: params.generatedAt, entries });
+}

extensions/qa-lab/src/gateway-child.ts

@@ -25,6 +25,7 @@ import {
   resolveQaRuntimeHostVersion,
 } from "./bundled-plugin-staging.js";
 import { assertRepoBoundPath, ensureRepoBoundDirectory } from "./cli-paths.js";
+import { QaSuiteInfraError } from "./errors.js";
 import { formatQaGatewayLogsForError, redactQaGatewayDebugText } from "./gateway-log-redaction.js";
 import { startQaGatewayRpcClient } from "./gateway-rpc-client.js";
 import { splitQaModelRef, type QaProviderMode } from "./model-selection.js";
@@ -470,7 +471,8 @@ async function waitForGatewayReady(params: {
   const startedAt = Date.now();
   while (Date.now() - startedAt < (params.timeoutMs ?? 60_000)) {
     if (params.child.exitCode !== null || params.child.signalCode !== null) {
-      throw new Error(
+      throw new QaSuiteInfraError(
+        "gateway_startup_unhealthy",
         `gateway exited before becoming healthy (exitCode=${String(params.child.exitCode)}, signal=${String(params.child.signalCode)}):\n${params.logs()}`,
       );
     }
@@ -485,7 +487,10 @@ async function waitForGatewayReady(params: {
     }
     await sleep(250);
   }
-  throw new Error(`gateway failed to become healthy:\n${params.logs()}`);
+  throw new QaSuiteInfraError(
+    "gateway_startup_unhealthy",
+    `gateway failed to become healthy:\n${params.logs()}`,
+  );
 }
 
 function isRetryableRpcStartupError(error: unknown) {
@@ -1031,14 +1036,13 @@ export async function startQaGatewayChild(params: {
         stagedBundledPluginsRoot,
       });
     }
-    throw new Error(
-      keepTemp
-        ? appendQaGatewayTempRoot(formatErrorMessage(error), tempRoot)
-        : formatErrorMessage(error),
-      {
-        cause: error,
-      },
-    );
+    const message = keepTemp
+      ? appendQaGatewayTempRoot(formatErrorMessage(error), tempRoot)
+      : formatErrorMessage(error);
+    if (error instanceof QaSuiteInfraError) {
+      throw new QaSuiteInfraError(error.code, message, { cause: error });
+    }
+    throw new Error(message, { cause: error });
   }
 }
 export { testing as __testing };

extensions/qa-lab/src/live-transports/discord/discord-live.runtime.ts

@@ -15,6 +15,7 @@ import { writeExternalFileWithinRoot } from "openclaw/plugin-sdk/security-runtim
 import { uniqueStrings } from "openclaw/plugin-sdk/string-coerce-runtime";
 import { chromium } from "playwright-core";
 import { z } from "zod";
+import { QA_EVIDENCE_FILENAME, buildLiveTransportEvidenceSummary } from "../../evidence-summary.js";
 import { startQaGatewayChild } from "../../gateway-child.js";
 import { DEFAULT_QA_LIVE_PROVIDER_MODE } from "../../providers/index.js";
 import {
@@ -25,7 +26,6 @@ import {
 import {
   acquireQaCredentialLease,
   startQaCredentialLeaseHeartbeat,
-  type QaCredentialRole,
 } from "../shared/credential-lease.runtime.js";
 import {
   appendQaLiveLaneIssue as appendLiveLaneIssue,
@@ -185,6 +185,7 @@ type DiscordQaScenarioResult = {
   artifactPaths?: Record<string, string>;
   id: string;
   title: string;
+  standardId?: string;
   status: "pass" | "fail";
   details: string;
   requestStartedAt?: string;
@@ -208,33 +209,6 @@ type DiscordQaRunResult = {
   scenarios: DiscordQaScenarioResult[];
 };
 
-type DiscordQaSummary = {
-  artifacts: {
-    observedMessagesPath: string;
-    reactionTimelinesPath?: string;
-    reportPath: string;
-    summaryPath: string;
-  };
-  credentials: {
-    credentialId?: string;
-    kind: string;
-    ownerId?: string;
-    role?: QaCredentialRole;
-    source: "convex" | "env";
-  };
-  guildId: string;
-  channelId: string;
-  startedAt: string;
-  finishedAt: string;
-  cleanupIssues: string[];
-  counts: {
-    total: number;
-    passed: number;
-    failed: number;
-  };
-  scenarios: DiscordQaScenarioResult[];
-};
-
 type DiscordReactionSnapshot = {
   elapsedMs: number;
   observedAt: string;
@@ -1302,6 +1276,7 @@ async function runDiscordThreadReplyFilePathAttachmentScenario(params: {
     return {
       id: params.scenario.id,
       title: params.scenario.title,
+      standardId: params.scenario.standardId,
       status,
       details:
         status === "pass"
@@ -1691,6 +1666,7 @@ export async function runDiscordQaLive(params: {
             scenarioResults.push({
               id: scenario.id,
               title: scenario.title,
+              standardId: scenario.standardId,
               status: "pass",
               details: redactPublicMetadata
                 ? "native command registered"
@@ -1712,6 +1688,7 @@ export async function runDiscordQaLive(params: {
             scenarioResults.push({
               id: scenario.id,
               title: scenario.title,
+              standardId: scenario.standardId,
               status: "pass",
               details: redactPublicMetadata
                 ? "SUT bot joined voice channel"
@@ -1766,6 +1743,7 @@ export async function runDiscordQaLive(params: {
             scenarioResults.push({
               id: scenario.id,
               title: scenario.title,
+              standardId: scenario.standardId,
               status: missing.length === 0 ? "pass" : "fail",
               details:
                 missing.length === 0
@@ -1811,6 +1789,7 @@ export async function runDiscordQaLive(params: {
           scenarioResults.push({
             id: scenario.id,
             title: scenario.title,
+            standardId: scenario.standardId,
             status: "pass",
             details: redactPublicMetadata
               ? "reply matched"
@@ -1838,6 +1817,7 @@ export async function runDiscordQaLive(params: {
               scenarioResults.push({
                 id: scenario.id,
                 title: scenario.title,
+                standardId: scenario.standardId,
                 status: "pass",
                 details: "no reply",
               });
@@ -1847,6 +1827,7 @@ export async function runDiscordQaLive(params: {
           scenarioResults.push({
             id: scenario.id,
             title: scenario.title,
+            standardId: scenario.standardId,
             status: "fail",
             details: formatErrorMessage(error),
           });
@@ -1879,40 +1860,26 @@ export async function runDiscordQaLive(params: {
   const publishedCleanupIssues = redactPublicMetadata
     ? redactQaLiveLaneIssues(cleanupIssues)
     : cleanupIssues;
-  const passedCount = scenarioResults.filter((entry) => entry.status === "pass").length;
-  const failedCount = scenarioResults.filter((entry) => entry.status === "fail").length;
-  const summary: DiscordQaSummary = {
-    artifacts: {
-      reportPath: path.join(outputDir, "discord-qa-report.md"),
-      summaryPath: path.join(outputDir, "discord-qa-summary.json"),
-      observedMessagesPath: path.join(outputDir, "discord-qa-observed-messages.json"),
-      ...(reactionTimelines.length > 0
-        ? { reactionTimelinesPath: path.join(outputDir, "discord-qa-reaction-timelines.json") }
-        : {}),
-    },
-    credentials: {
-      source: credentialLease.source,
-      kind: credentialLease.kind,
-      role: credentialLease.role,
-      ownerId: redactPublicMetadata ? undefined : credentialLease.ownerId,
-      credentialId: redactPublicMetadata ? undefined : credentialLease.credentialId,
-    },
-    guildId: redactPublicMetadata ? "<redacted>" : runtimeEnv.guildId,
-    channelId: redactPublicMetadata ? "<redacted>" : runtimeEnv.channelId,
-    startedAt,
-    finishedAt,
-    cleanupIssues: publishedCleanupIssues,
-    counts: {
-      total: scenarioResults.length,
-      passed: passedCount,
-      failed: failedCount,
-    },
-    scenarios: scenarioResults,
-  };
   const reportPath = path.join(outputDir, "discord-qa-report.md");
-  const summaryPath = path.join(outputDir, "discord-qa-summary.json");
+  const summaryPath = path.join(outputDir, QA_EVIDENCE_FILENAME);
   const observedMessagesPath = path.join(outputDir, "discord-qa-observed-messages.json");
   const reactionTimelinesPath = path.join(outputDir, "discord-qa-reaction-timelines.json");
+  const evidence = buildLiveTransportEvidenceSummary({
+    artifactPaths: [
+      { kind: "summary", path: path.basename(summaryPath) },
+      { kind: "report", path: path.basename(reportPath) },
+      { kind: "transport-observations", path: path.basename(observedMessagesPath) },
+      ...(reactionTimelines.length > 0
+        ? [{ kind: "reaction-timelines", path: path.basename(reactionTimelinesPath) }]
+        : []),
+    ],
+    checks: scenarioResults,
+    env: process.env,
+    generatedAt: finishedAt,
+    primaryModel,
+    providerMode,
+    transportId: "discord",
+  });
   await fs.writeFile(
     reportPath,
     `${renderDiscordQaMarkdown({
@@ -1928,7 +1895,7 @@ export async function runDiscordQaLive(params: {
     })}\n`,
     { encoding: "utf8", mode: 0o600 },
   );
-  await fs.writeFile(summaryPath, `${JSON.stringify(summary, null, 2)}\n`, {
+  await fs.writeFile(summaryPath, `${JSON.stringify(evidence, null, 2)}\n`, {
     encoding: "utf8",
     mode: 0o600,
   });

extensions/qa-lab/src/live-transports/slack/slack-live.runtime.test.ts

@@ -3,6 +3,7 @@ import fs from "node:fs/promises";
 import { tmpdir } from "node:os";
 import path from "node:path";
 import { beforeEach, describe, expect, it, vi } from "vitest";
+import { QA_EVIDENCE_FILENAME, QA_EVIDENCE_SUMMARY_KIND } from "../../evidence-summary.js";
 import { testing, runSlackQaLive } from "./slack-live.runtime.js";
 
 describe("Slack live QA runtime helpers", () => {
@@ -53,7 +54,7 @@ describe("Slack live QA runtime helpers", () => {
     });
   });
 
-  it("reports standard live transport scenario coverage", () => {
+  it("reports live transport standard scenario coverage", () => {
     expect(testing.SLACK_QA_STANDARD_SCENARIO_IDS).toEqual([
       "canary",
       "mention-gating",
@@ -71,7 +72,7 @@ describe("Slack live QA runtime helpers", () => {
     ]);
   });
 
-  it("selects native approval scenarios by id without changing standard coverage", () => {
+  it("selects native approval scenarios by id without changing standard scenario coverage", () => {
     expect(
       testing
         .findScenario(["slack-approval-exec-native", "slack-approval-plugin-native"])
@@ -436,15 +437,25 @@ describe("Slack live QA runtime helpers", () => {
     expect(result.scenarios[0]?.status).toBe("fail");
     expect(result.scenarios[0]?.details).toContain("Missing OPENCLAW_QA_CONVEX_SITE_URL");
     await expect(fs.stat(result.reportPath).then((stats) => stats.isFile())).resolves.toBe(true);
+    expect(path.basename(result.summaryPath)).toBe(QA_EVIDENCE_FILENAME);
     const summary = JSON.parse(await fs.readFile(result.summaryPath, "utf8")) as {
-      channelId: string;
-      credentials: { kind: string; role?: string; source: string };
+      entries: Array<{
+        result: { failure?: { reason?: string }; status: string };
+        test: { id: string };
+      }>;
+      kind: string;
     };
-    expect(summary.channelId).toBe("<unavailable>");
-    expect(summary.credentials).toEqual({
-      kind: "slack",
-      role: "ci",
-      source: "convex",
+    expect(summary.kind).toBe(QA_EVIDENCE_SUMMARY_KIND);
+    expect(summary.entries[0]).toMatchObject({
+      test: {
+        id: "slack-canary",
+      },
+      result: {
+        status: "fail",
+        failure: {
+          reason: expect.stringContaining("Missing OPENCLAW_QA_CONVEX_SITE_URL"),
+        },
+      },
     });
   });
 });

extensions/qa-lab/src/live-transports/slack/slack-live.runtime.ts

@@ -9,6 +9,7 @@ import { formatErrorMessage } from "openclaw/plugin-sdk/error-runtime";
 import { parseStrictPositiveInteger } from "openclaw/plugin-sdk/number-runtime";
 import { uniqueStrings } from "openclaw/plugin-sdk/string-coerce-runtime";
 import { z } from "zod";
+import { QA_EVIDENCE_FILENAME, buildLiveTransportEvidenceSummary } from "../../evidence-summary.js";
 import { startQaGatewayChild } from "../../gateway-child.js";
 import { DEFAULT_QA_LIVE_PROVIDER_MODE } from "../../providers/index.js";
 import {
@@ -19,7 +20,6 @@ import {
 import {
   acquireQaCredentialLease,
   startQaCredentialLeaseHeartbeat,
-  type QaCredentialRole,
 } from "../shared/credential-lease.runtime.js";
 import {
   appendQaLiveLaneIssue as appendLiveLaneIssue,
@@ -215,6 +215,7 @@ type SlackQaScenarioResult = {
     responseObservedAt: string;
     source: "approval-request-to-resolution" | "request-to-observed-message";
   };
+  standardId?: string;
   status: "fail" | "pass";
   title: string;
 };
@@ -228,26 +229,6 @@ export type SlackQaRunResult = {
   summaryPath: string;
 };
 
-type SlackQaSummary = {
-  channelId: string;
-  cleanupIssues: string[];
-  counts: {
-    failed: number;
-    passed: number;
-    total: number;
-  };
-  credentials: {
-    credentialId?: string;
-    kind: string;
-    ownerId?: string;
-    role?: QaCredentialRole;
-    source: "convex" | "env";
-  };
-  finishedAt: string;
-  scenarios: SlackQaScenarioResult[];
-  startedAt: string;
-};
-
 type SlackCredentialLease = Awaited<ReturnType<typeof acquireQaCredentialLease<SlackQaRuntimeEnv>>>;
 type SlackCredentialHeartbeat = ReturnType<typeof startQaCredentialLeaseHeartbeat>;
 
@@ -537,14 +518,6 @@ function inferSlackCredentialSource(
   return normalized === "convex" ? "convex" : "env";
 }
 
-function inferSlackCredentialRole(value: string | undefined): QaCredentialRole | undefined {
-  const normalized = value?.trim().toLowerCase();
-  if (normalized === "ci" || normalized === "maintainer") {
-    return normalized;
-  }
-  return undefined;
-}
-
 function normalizeSlackId(value: string, label: string) {
   const normalized = value.trim();
   if (!/^[A-Z][A-Z0-9]+$/.test(normalized)) {
@@ -1804,7 +1777,6 @@ export async function runSlackQaLive(params: {
   const sutAccountId = params.sutAccountId?.trim() || "sut";
   const scenarios = findScenario(params.scenarioIds);
   const requestedCredentialSource = inferSlackCredentialSource(params.credentialSource);
-  const requestedCredentialRole = inferSlackCredentialRole(params.credentialRole);
   const redactPublicMetadata = isTruthyOptIn(process.env[QA_REDACT_PUBLIC_METADATA_ENV]);
   const includeObservedMessageContent = isTruthyOptIn(process.env[SLACK_QA_CAPTURE_CONTENT_ENV]);
   const startedAt = new Date().toISOString();
@@ -1916,6 +1888,7 @@ export async function runSlackQaLive(params: {
               approval: approval.artifact,
               id: scenario.id,
               title: scenario.title,
+              standardId: scenario.standardId,
               status: "pass",
               details: [
                 `${scenarioRun.approvalKind} approval resolved ${scenarioRun.decision} in ${approval.rttMs}ms`,
@@ -1972,6 +1945,7 @@ export async function runSlackQaLive(params: {
             scenarioResults.push({
               id: scenario.id,
               title: scenario.title,
+              standardId: scenario.standardId,
               status: "pass",
               details: [
                 `reply matched in ${rttMs}ms`,
@@ -2006,6 +1980,7 @@ export async function runSlackQaLive(params: {
             scenarioResults.push({
               id: scenario.id,
               title: scenario.title,
+              standardId: scenario.standardId,
               status: "pass",
               details:
                 scenarioAttempt > 1 ? `no reply; retried ${scenarioAttempt - 1}x` : "no reply",
@@ -2023,6 +1998,7 @@ export async function runSlackQaLive(params: {
           scenarioResults.push({
             id: scenario.id,
             title: scenario.title,
+            standardId: scenario.standardId,
             status: "fail",
             details:
               scenarioAttempt > 1
@@ -2071,6 +2047,7 @@ export async function runSlackQaLive(params: {
     scenarioResults.push({
       id: "slack-canary",
       title: "Slack canary echo",
+      standardId: "canary",
       status: "fail",
       details: formatErrorMessage(error),
     });
@@ -2093,44 +2070,26 @@ export async function runSlackQaLive(params: {
 
   const finishedAt = new Date().toISOString();
   const reportPath = path.join(outputDir, "slack-qa-report.md");
-  const summaryPath = path.join(outputDir, "slack-qa-summary.json");
+  const summaryPath = path.join(outputDir, QA_EVIDENCE_FILENAME);
   const observedMessagesPath = path.join(outputDir, "slack-qa-observed-messages.json");
-  const passed = scenarioResults.filter((entry) => entry.status === "pass").length;
-  const failed = scenarioResults.filter((entry) => entry.status === "fail").length;
   const artifactScenarioResults = toSlackQaScenarioArtifactResults({
     scenarios: scenarioResults,
     includeContent: includeObservedMessageContent,
     redactMetadata: redactPublicMetadata,
   });
-  const summary: SlackQaSummary = {
-    credentials: credentialLease
-      ? {
-          source: credentialLease.source,
-          kind: credentialLease.kind,
-          role: credentialLease.role,
-          credentialId: redactPublicMetadata ? undefined : credentialLease.credentialId,
-          ownerId: redactPublicMetadata ? undefined : credentialLease.ownerId,
-        }
-      : {
-          source: requestedCredentialSource,
-          kind: "slack",
-          role: requestedCredentialRole,
-        },
-    channelId: runtimeEnv
-      ? redactPublicMetadata
-        ? "<redacted>"
-        : runtimeEnv.channelId
-      : "<unavailable>",
-    startedAt,
-    finishedAt,
-    cleanupIssues,
-    counts: {
-      total: scenarioResults.length,
-      passed,
-      failed,
-    },
-    scenarios: artifactScenarioResults,
-  };
+  const evidence = buildLiveTransportEvidenceSummary({
+    artifactPaths: [
+      { kind: "summary", path: path.basename(summaryPath) },
+      { kind: "report", path: path.basename(reportPath) },
+      { kind: "transport-observations", path: path.basename(observedMessagesPath) },
+    ],
+    checks: artifactScenarioResults,
+    env: process.env,
+    generatedAt: finishedAt,
+    primaryModel,
+    providerMode,
+    transportId: "slack",
+  });
   await fs.writeFile(
     observedMessagesPath,
     `${JSON.stringify(
@@ -2143,7 +2102,7 @@ export async function runSlackQaLive(params: {
       2,
     )}\n`,
   );
-  await fs.writeFile(summaryPath, `${JSON.stringify(summary, null, 2)}\n`);
+  await fs.writeFile(summaryPath, `${JSON.stringify(evidence, null, 2)}\n`);
   await fs.writeFile(
     reportPath,
     `${renderSlackQaMarkdown({

extensions/qa-lab/src/live-transports/telegram/telegram-live.runtime.ts

@@ -14,6 +14,7 @@ import { fetchWithSsrFGuard } from "openclaw/plugin-sdk/ssrf-runtime";
 import { isRecord, uniqueStrings } from "openclaw/plugin-sdk/string-coerce-runtime";
 import { resolvePreferredOpenClawTmpDir } from "openclaw/plugin-sdk/temp-path";
 import { z } from "zod";
+import { QA_EVIDENCE_FILENAME, buildLiveTransportEvidenceSummary } from "../../evidence-summary.js";
 import { startQaGatewayChild } from "../../gateway-child.js";
 import { DEFAULT_QA_LIVE_PROVIDER_MODE } from "../../providers/index.js";
 import {
@@ -25,7 +26,6 @@ import {
 import {
   acquireQaCredentialLease,
   startQaCredentialLeaseHeartbeat,
-  type QaCredentialRole,
 } from "../shared/credential-lease.runtime.js";
 import {
   appendQaLiveLaneIssue as appendLiveLaneIssue,
@@ -138,6 +138,7 @@ const DEFAULT_TELEGRAM_QA_CANARY_TIMEOUT_MS = 30_000;
 
 type TelegramQaScenarioResult = {
   id: string;
+  standardId?: string;
   title: string;
   status: "pass" | "fail";
   details: string;
@@ -165,26 +166,6 @@ type TelegramQaRunResult = {
   scenarios: TelegramQaScenarioResult[];
 };
 
-type TelegramQaSummary = {
-  credentials: {
-    credentialId?: string;
-    kind: string;
-    ownerId?: string;
-    role?: QaCredentialRole;
-    source: "convex" | "env";
-  };
-  groupId: string;
-  startedAt: string;
-  finishedAt: string;
-  cleanupIssues: string[];
-  counts: {
-    total: number;
-    passed: number;
-    failed: number;
-  };
-  scenarios: TelegramQaScenarioResult[];
-};
-
 class TelegramQaCanaryError extends Error {
   phase: TelegramQaCanaryPhase;
   context: Record<string, string | number | undefined>;
@@ -1785,6 +1766,7 @@ export async function runTelegramQaLive(params: {
         latestSutMessageId = canaryTiming.responseMessageId;
         scenarioResults.push({
           id: "telegram-canary",
+          standardId: "canary",
           title: "Telegram canary",
           status: "pass",
           details: redactPublicMetadata
@@ -1815,6 +1797,7 @@ export async function runTelegramQaLive(params: {
         });
         scenarioResults.push({
           id: "telegram-canary",
+          standardId: "canary",
           title: "Telegram canary",
           status: "fail",
           details: canaryFailure,
@@ -1910,6 +1893,7 @@ export async function runTelegramQaLive(params: {
             if (!lastMatched || !firstRequestStartedAt || lastSentMessageId === undefined) {
               const result = {
                 id: scenario.id,
+                standardId: scenario.standardId,
                 title: scenario.title,
                 status: "pass",
                 details: "no reply",
@@ -1933,6 +1917,7 @@ export async function runTelegramQaLive(params: {
                 : `; ${scenarioSteps.filter((step) => step.expectReply).length} command replies matched`;
             const result = {
               id: scenario.id,
+              standardId: scenario.standardId,
               title: scenario.title,
               status: "pass",
               details: redactPublicMetadata
@@ -1958,6 +1943,7 @@ export async function runTelegramQaLive(params: {
           } catch (error) {
             const result = {
               id: scenario.id,
+              standardId: scenario.standardId,
               title: scenario.title,
               status: "fail",
               details: formatErrorMessage(error),
@@ -2006,28 +1992,22 @@ export async function runTelegramQaLive(params: {
   if (cleanupIssues.length > 0) {
     writeTelegramQaProgress(progressEnabled, `cleanup issues: count=${cleanupIssues.length}`);
   }
-  const summary: TelegramQaSummary = {
-    credentials: {
-      source: credentialLease.source,
-      kind: credentialLease.kind,
-      role: credentialLease.role,
-      ownerId: redactPublicMetadata ? undefined : credentialLease.ownerId,
-      credentialId: redactPublicMetadata ? undefined : credentialLease.credentialId,
-    },
-    groupId: redactPublicMetadata ? "<redacted>" : runtimeEnv.groupId,
-    startedAt,
-    finishedAt,
-    cleanupIssues: publishedCleanupIssues,
-    counts: {
-      total: scenarioResults.length,
-      passed: passedCount,
-      failed: failedCount,
-    },
-    scenarios: scenarioResults,
-  };
   const reportPath = path.join(outputDir, "telegram-qa-report.md");
-  const summaryPath = path.join(outputDir, "telegram-qa-summary.json");
+  const summaryPath = path.join(outputDir, QA_EVIDENCE_FILENAME);
   const observedMessagesPath = path.join(outputDir, "telegram-qa-observed-messages.json");
+  const evidence = buildLiveTransportEvidenceSummary({
+    artifactPaths: [
+      { kind: "summary", path: path.basename(summaryPath) },
+      { kind: "report", path: path.basename(reportPath) },
+      { kind: "transport-observations", path: path.basename(observedMessagesPath) },
+    ],
+    env: process.env,
+    generatedAt: finishedAt,
+    primaryModel,
+    providerMode,
+    checks: scenarioResults,
+    transportId: "telegram",
+  });
   await fs.writeFile(
     reportPath,
     `${renderTelegramQaMarkdown({
@@ -2042,7 +2022,7 @@ export async function runTelegramQaLive(params: {
     })}\n`,
     { encoding: "utf8", mode: 0o600 },
   );
-  await fs.writeFile(summaryPath, `${JSON.stringify(summary, null, 2)}\n`, {
+  await fs.writeFile(summaryPath, `${JSON.stringify(evidence, null, 2)}\n`, {
     encoding: "utf8",
     mode: 0o600,
   });

extensions/qa-lab/src/live-transports/whatsapp/cli.runtime.test.ts

@@ -3,6 +3,7 @@ import fs from "node:fs/promises";
 import os from "node:os";
 import path from "node:path";
 import { afterEach, describe, expect, it, vi } from "vitest";
+import { QA_EVIDENCE_FILENAME } from "../../evidence-summary.js";
 import { runQaWhatsAppCommand } from "./cli.runtime.js";
 
 const runWhatsAppQaLiveMock = vi.hoisted(() => vi.fn());
@@ -38,24 +39,48 @@ afterEach(async () => {
 async function writeSummary(summary: unknown) {
   const outputDir = await fs.mkdtemp(path.join(os.tmpdir(), "openclaw-whatsapp-cli-"));
   tempDirs.push(outputDir);
-  const summaryPath = path.join(outputDir, "whatsapp-qa-summary.json");
+  const summaryPath = path.join(outputDir, QA_EVIDENCE_FILENAME);
   await fs.writeFile(summaryPath, `${JSON.stringify(summary, null, 2)}\n`, "utf8");
   return { outputDir, summaryPath };
 }
 
+function makeEvidenceSummary(status: "pass" | "fail" | "blocked" | "skipped") {
+  return {
+    kind: "openclaw.qa.evidence-summary",
+    schemaVersion: 2,
+    generatedAt: "2026-05-01T00:00:00.000Z",
+    entries: [
+      {
+        test: {
+          kind: "live-transport-check",
+          id: "whatsapp-mention-gating",
+          title: "WhatsApp mention gating",
+        },
+        mapping: { profile: "release", coverage: [] },
+        execution: {
+          runner: "host",
+          environment: { ref: null, os: "darwin", nodeVersion: "v24.0.0" },
+          provider: {
+            id: "openai",
+            live: false,
+            model: { name: null, ref: null },
+            fixture: "mock-openai",
+          },
+          channel: { id: "whatsapp", live: true, driver: "native" },
+          packageSource: { kind: "source-checkout" },
+          artifacts: [],
+        },
+        result: { status },
+      },
+    ],
+  };
+}
+
 describe("WhatsApp QA CLI runtime", () => {
-  it("fails when a standard scenario is skipped by default", async () => {
+  it("fails when a requirement is skipped by default", async () => {
     originalExitCode = process.exitCode;
     process.exitCode = undefined;
-    const { outputDir, summaryPath } = await writeSummary({
-      counts: { total: 1, passed: 0, failed: 0, skipped: 1 },
-      scenarios: [
-        {
-          id: "whatsapp-mention-gating",
-          status: "skip",
-        },
-      ],
-    });
+    const { outputDir, summaryPath } = await writeSummary(makeEvidenceSummary("skipped"));
     runWhatsAppQaLiveMock.mockResolvedValueOnce({
       observedMessagesPath: path.join(outputDir, "observed.json"),
       reportPath: path.join(outputDir, "report.md"),
@@ -71,10 +96,7 @@ describe("WhatsApp QA CLI runtime", () => {
   it("allows skipped scenarios when failures are explicitly allowed", async () => {
     originalExitCode = process.exitCode;
     process.exitCode = undefined;
-    const { outputDir, summaryPath } = await writeSummary({
-      counts: { total: 1, passed: 0, failed: 0, skipped: 1 },
-      scenarios: [{ id: "whatsapp-mention-gating", status: "skip" }],
-    });
+    const { outputDir, summaryPath } = await writeSummary(makeEvidenceSummary("skipped"));
     runWhatsAppQaLiveMock.mockResolvedValueOnce({
       observedMessagesPath: path.join(outputDir, "observed.json"),
       reportPath: path.join(outputDir, "report.md"),

extensions/qa-lab/src/live-transports/whatsapp/whatsapp-live.runtime.test.ts

@@ -332,7 +332,7 @@ describe("WhatsApp QA live runtime", () => {
     expect(scenarios.map(({ id }) => id)).toEqual(["whatsapp-canary", "whatsapp-pairing-block"]);
   });
 
-  it("reports standard WhatsApp live transport scenario coverage", () => {
+  it("reports WhatsApp live transport standard scenario coverage", () => {
     expect(testing.WHATSAPP_QA_STANDARD_SCENARIO_IDS).toEqual([
       "canary",
       "mention-gating",
@@ -904,7 +904,7 @@ describe("WhatsApp QA live runtime", () => {
     expect(waitCallCount).toBe(2);
   });
 
-  it("selects native approval scenarios by id without changing standard coverage", () => {
+  it("selects native approval scenarios by id without changing standard scenario coverage", () => {
     const scenarios = testing.findScenarios([
       "whatsapp-approval-exec-native",
       "whatsapp-approval-exec-reaction-native",
@@ -1181,6 +1181,7 @@ describe("WhatsApp QA live runtime", () => {
       {
         id: "whatsapp-mention-gating",
         title: "WhatsApp group mention gating",
+        standardId: "mention-gating",
         status: "fail",
         details: "setup exploded",
       },

extensions/qa-lab/src/live-transports/whatsapp/whatsapp-live.runtime.ts

@@ -15,6 +15,7 @@ import { formatErrorMessage } from "openclaw/plugin-sdk/error-runtime";
 import { normalizeStringEntries, uniqueStrings } from "openclaw/plugin-sdk/string-coerce-runtime";
 import { resolvePreferredOpenClawTmpDir } from "openclaw/plugin-sdk/temp-path";
 import { z } from "zod";
+import { QA_EVIDENCE_FILENAME, buildLiveTransportEvidenceSummary } from "../../evidence-summary.js";
 import { startQaGatewayChild } from "../../gateway-child.js";
 import { DEFAULT_QA_LIVE_PROVIDER_MODE } from "../../providers/index.js";
 import { fingerprintQaCredentialId } from "../../qa-credentials-fingerprint.runtime.js";
@@ -27,7 +28,6 @@ import {
 import {
   acquireQaCredentialLease,
   startQaCredentialLeaseHeartbeat,
-  type QaCredentialRole,
 } from "../shared/credential-lease.runtime.js";
 import {
   appendQaLiveLaneIssue as appendLiveLaneIssue,
@@ -258,6 +258,7 @@ type WhatsAppQaScenarioResult = {
     responseObservedAt: string;
     source: "approval-request-to-resolution" | "request-to-observed-message";
   };
+  standardId?: string;
   status: "fail" | "pass" | "skip";
   title: string;
 };
@@ -271,29 +272,6 @@ export type WhatsAppQaRunResult = {
   summaryPath: string;
 };
 
-type WhatsAppQaSummary = {
-  cleanupIssues: string[];
-  counts: {
-    failed: number;
-    passed: number;
-    skipped: number;
-    total: number;
-  };
-  credentials: {
-    credentialFingerprint?: string;
-    credentialId?: string;
-    kind: string;
-    ownerId?: string;
-    role?: QaCredentialRole;
-    source: "convex" | "env";
-  };
-  finishedAt: string;
-  scenarios: WhatsAppQaScenarioResult[];
-  startedAt: string;
-  sutAccountId: string;
-  sutPhoneE164: string;
-};
-
 type WhatsAppCredentialLease = Awaited<
   ReturnType<typeof acquireQaCredentialLease<WhatsAppQaRuntimeEnv>>
 >;
@@ -1427,14 +1405,6 @@ function inferWhatsAppCredentialSource(
   return normalized === "convex" ? "convex" : "env";
 }
 
-function inferWhatsAppCredentialRole(value: string | undefined): QaCredentialRole | undefined {
-  const normalized = value?.trim().toLowerCase();
-  if (normalized === "ci" || normalized === "maintainer") {
-    return normalized;
-  }
-  return undefined;
-}
-
 function resolveWhatsAppMetadataRedaction(env: NodeJS.ProcessEnv = process.env) {
   const raw = env[QA_REDACT_PUBLIC_METADATA_ENV];
   return raw === undefined ? true : isTruthyOptIn(raw);
@@ -2630,6 +2600,7 @@ async function runWhatsAppScenario(params: {
       return {
         id: params.scenario.id,
         title: params.scenario.title,
+        standardId: params.scenario.standardId,
         status: "pass" as const,
         details: `${scenarioRun.approvalKind} approval ${approval.approvalId} resolved ${scenarioRun.decision} in ${approval.rttMs}ms`,
         rttMs: approval.rttMs,
@@ -2733,6 +2704,7 @@ async function runWhatsAppScenario(params: {
       return {
         id: params.scenario.id,
         title: params.scenario.title,
+        standardId: params.scenario.standardId,
         status: "pass" as const,
         details: "no reply",
       };
@@ -2766,6 +2738,7 @@ async function runWhatsAppScenario(params: {
     return {
       id: params.scenario.id,
       title: params.scenario.title,
+      standardId: params.scenario.standardId,
       status: "pass" as const,
       details: [`reply matched in ${rttMs}ms`, afterSendDetails, afterReplyDetails, batchDetails]
         .filter(Boolean)
@@ -2945,6 +2918,7 @@ function createMissingGroupJidScenarioResult(params: {
   return {
     id: params.scenario.id,
     title: params.scenario.title,
+    standardId: params.scenario.standardId,
     status: params.explicitScenarioSelection ? "fail" : "skip",
     details: params.explicitScenarioSelection
       ? "requested scenario requires groupJid in the WhatsApp QA credential payload"
@@ -2967,6 +2941,7 @@ function appendPreScenarioFailureResults(params: {
     params.scenarioResults.push({
       id: scenario.id,
       title: scenario.title,
+      standardId: scenario.standardId,
       status: "fail",
       details: params.details,
     });
@@ -3025,7 +3000,6 @@ export async function runWhatsAppQaLive(params: {
   const scenarios = findScenarios(params.scenarioIds, providerMode);
   const explicitScenarioSelection = (params.scenarioIds?.length ?? 0) > 0;
   const requestedCredentialSource = inferWhatsAppCredentialSource(params.credentialSource);
-  const requestedCredentialRole = inferWhatsAppCredentialRole(params.credentialRole);
   const redactPublicMetadata = resolveWhatsAppMetadataRedaction();
   const includeObservedMessageContent = isTruthyOptIn(process.env[WHATSAPP_QA_CAPTURE_CONTENT_ENV]);
   const startedAt = new Date().toISOString();
@@ -3156,6 +3130,7 @@ export async function runWhatsAppQaLive(params: {
           const result: WhatsAppQaScenarioResult = {
             id: scenario.id,
             title: scenario.title,
+            standardId: scenario.standardId,
             status: "fail",
             details:
               driverAttempt > 1
@@ -3228,11 +3203,8 @@ export async function runWhatsAppQaLive(params: {
 
   const finishedAt = new Date().toISOString();
   const reportPath = path.join(outputDir, "whatsapp-qa-report.md");
-  const summaryPath = path.join(outputDir, "whatsapp-qa-summary.json");
+  const summaryPath = path.join(outputDir, QA_EVIDENCE_FILENAME);
   const observedMessagesPath = path.join(outputDir, "whatsapp-qa-observed-messages.json");
-  const passed = scenarioResults.filter((entry) => entry.status === "pass").length;
-  const failed = scenarioResults.filter((entry) => entry.status === "fail").length;
-  const skipped = scenarioResults.filter((entry) => entry.status === "skip").length;
   const credentialFingerprint = fingerprintQaCredentialId(credentialLease?.credentialId);
   const publishedCleanupIssues = redactPublicMetadata
     ? redactQaLiveLaneIssues(cleanupIssues)
@@ -3240,36 +3212,19 @@ export async function runWhatsAppQaLive(params: {
   const publishedScenarioResults = redactPublicMetadata
     ? redactWhatsAppQaScenarioResults(scenarioResults)
     : scenarioResults;
-  const summary: WhatsAppQaSummary = {
-    credentials: credentialLease
-      ? {
-          source: credentialLease.source,
-          kind: credentialLease.kind,
-          role: credentialLease.role,
-          credentialFingerprint,
-          credentialId: redactPublicMetadata ? undefined : credentialLease.credentialId,
-          ownerId: redactPublicMetadata ? undefined : credentialLease.ownerId,
-        }
-      : {
-          source: requestedCredentialSource,
-          kind: "whatsapp",
-          role: requestedCredentialRole,
-        },
-    sutAccountId,
-    sutPhoneE164: redactPublicMetadata
-      ? "<redacted>"
-      : (runtimeEnv?.sutPhoneE164 ?? "<unavailable>"),
-    startedAt,
-    finishedAt,
-    cleanupIssues: publishedCleanupIssues,
-    counts: {
-      total: scenarioResults.length,
-      passed,
-      failed,
-      skipped,
-    },
-    scenarios: publishedScenarioResults,
-  };
+  const evidence = buildLiveTransportEvidenceSummary({
+    artifactPaths: [
+      { kind: "summary", path: path.basename(summaryPath) },
+      { kind: "report", path: path.basename(reportPath) },
+      { kind: "transport-observations", path: path.basename(observedMessagesPath) },
+    ],
+    checks: publishedScenarioResults,
+    env: process.env,
+    generatedAt: finishedAt,
+    primaryModel,
+    providerMode,
+    transportId: "whatsapp",
+  });
   await fs.writeFile(
     observedMessagesPath,
     `${JSON.stringify(
@@ -3282,7 +3237,7 @@ export async function runWhatsAppQaLive(params: {
       2,
     )}\n`,
   );
-  await fs.writeFile(summaryPath, `${JSON.stringify(summary, null, 2)}\n`);
+  await fs.writeFile(summaryPath, `${JSON.stringify(evidence, null, 2)}\n`);
   await fs.writeFile(
     reportPath,
     `${renderWhatsAppQaMarkdown({

extensions/qa-lab/src/mantis/run.runtime.test.ts

@@ -3,6 +3,7 @@ import fs from "node:fs/promises";
 import os from "node:os";
 import path from "node:path";
 import { afterEach, beforeEach, describe, expect, it, vi } from "vitest";
+import { QA_EVIDENCE_FILENAME, buildLiveTransportEvidenceSummary } from "../evidence-summary.js";
 import { runMantisBeforeAfter } from "./run.runtime.js";
 
 describe("mantis before/after runtime", () => {
@@ -32,25 +33,31 @@ describe("mantis before/after runtime", () => {
       const videoPath = path.join(outputDir, `${lane}-timeline.mp4`);
       await fs.writeFile(screenshotPath, `${lane} screenshot`);
       await fs.writeFile(videoPath, `${lane} video`);
-      await fs.writeFile(
-        path.join(outputDir, "discord-qa-summary.json"),
-        `${JSON.stringify(
+      const summary = buildLiveTransportEvidenceSummary({
+        artifactPaths: [
+          { kind: "summary", path: QA_EVIDENCE_FILENAME },
+          { kind: "report", path: "discord-qa-report.md" },
+        ],
+        checks: [
           {
-            scenarios: [
-              {
-                artifactPaths: { screenshot: screenshotPath, video: videoPath },
-                details:
-                  lane === "baseline"
-                    ? "reaction timeline missing thinking/done"
-                    : "reaction timeline matched queued -> thinking -> done",
-                id: "discord-status-reactions-tool-only",
-                status: lane === "baseline" ? "fail" : "pass",
-              },
-            ],
+            artifactPaths: { screenshot: screenshotPath, video: videoPath },
+            details:
+              lane === "baseline"
+                ? "reaction timeline missing thinking/done"
+                : "reaction timeline matched queued -> thinking -> done",
+            id: "discord-status-reactions-tool-only",
+            status: lane === "baseline" ? "fail" : "pass",
+            title: "Discord explicit status reactions run in tool-only reply mode",
           },
-          null,
-          2,
-        )}\n`,
+        ],
+        generatedAt: "2026-05-03T12:00:00.000Z",
+        primaryModel: "openai/gpt-5.4",
+        providerMode: "live-frontier",
+        transportId: "discord",
+      });
+      await fs.writeFile(
+        path.join(outputDir, QA_EVIDENCE_FILENAME),
+        `${JSON.stringify(summary, null, 2)}\n`,
       );
     });
 

extensions/qa-lab/src/mantis/run.runtime.ts

@@ -4,6 +4,7 @@ import fs from "node:fs/promises";
 import path from "node:path";
 import { formatErrorMessage } from "openclaw/plugin-sdk/error-runtime";
 import { ensureRepoBoundDirectory, resolveRepoRelativeOutputDir } from "../cli-paths.js";
+import { QA_EVIDENCE_FILENAME, validateQaEvidenceSummaryJson } from "../evidence-summary.js";
 
 export type MantisBeforeAfterOptions = {
   allowFailures?: boolean;
@@ -47,6 +48,14 @@ type DiscordQaSummary = {
   }[];
 };
 
+type NormalizedScenarioSummary = {
+  details?: string;
+  screenshotPath?: string;
+  status: string;
+  summaryPath: string;
+  videoPath?: string;
+};
+
 type LaneResult = {
   outputDir: string;
   scenarioDetails?: string;
@@ -194,6 +203,18 @@ async function readLaneResult(params: {
   publishedLaneDir: string;
   scenario: string;
 }) {
+  const normalized = await readNormalizedLaneResult(params);
+  if (normalized) {
+    return {
+      outputDir: params.publishedLaneDir,
+      scenarioDetails: normalized.details,
+      screenshotPath: normalized.screenshotPath,
+      status: normalized.status,
+      summaryPath: normalized.summaryPath,
+      videoPath: normalized.videoPath,
+    } satisfies LaneResult;
+  }
+
   const summaryPath = path.join(params.publishedLaneDir, "discord-qa-summary.json");
   const summary = JSON.parse(await fs.readFile(summaryPath, "utf8")) as DiscordQaSummary;
   const scenarioSummary =
@@ -211,6 +232,35 @@ async function readLaneResult(params: {
   } satisfies LaneResult;
 }
 
+async function readNormalizedLaneResult(params: {
+  publishedLaneDir: string;
+  scenario: string;
+}): Promise<NormalizedScenarioSummary | undefined> {
+  const summaryPath = path.join(params.publishedLaneDir, QA_EVIDENCE_FILENAME);
+  let rawSummary: string;
+  try {
+    rawSummary = await fs.readFile(summaryPath, "utf8");
+  } catch (error) {
+    if ((error as NodeJS.ErrnoException).code === "ENOENT") {
+      return undefined;
+    }
+    throw error;
+  }
+
+  const summary = validateQaEvidenceSummaryJson(JSON.parse(rawSummary));
+  const entry =
+    summary.entries.find((candidate) => candidate.test.id === params.scenario) ??
+    summary.entries[0];
+  const artifacts = entry?.execution.artifacts ?? [];
+  return {
+    details: entry?.result.failure?.reason,
+    screenshotPath: artifacts.find((artifact) => artifact.kind === "screenshot")?.path,
+    status: entry?.result.status ?? "fail",
+    summaryPath,
+    videoPath: artifacts.find((artifact) => artifact.kind === "video")?.path,
+  };
+}
+
 function renderReport(params: {
   baseline: LaneResult;
   candidate: LaneResult;

extensions/qa-lab/src/qa-channel-transport.ts

@@ -3,6 +3,7 @@ import { setTimeout as sleep } from "node:timers/promises";
 import type { OpenClawConfig } from "openclaw/plugin-sdk/config-contracts";
 import { formatErrorMessage } from "openclaw/plugin-sdk/error-runtime";
 import type { QaBusState } from "./bus-state.js";
+import { QaSuiteInfraError } from "./errors.js";
 import { getQaProvider } from "./providers/index.js";
 import { QaStateBackedTransportAdapter } from "./qa-transport.js";
 import type {
@@ -65,7 +66,8 @@ async function waitForQaChannelReady(params: {
     await sleep(pollIntervalMs);
   }
 
-  throw new Error(
+  throw new QaSuiteInfraError(
+    "transport_ready_timeout",
     [
       `timed out after ${timeoutMs}ms waiting for qa-channel ready`,
       `last status: ${lastAccountStatus}`,

extensions/qa-lab/src/qa-gateway-config.test.ts

@@ -52,6 +52,11 @@ function getModelFallbacks(value: unknown): string[] | undefined {
   return undefined;
 }
 
+function expectQaLabPluginEnabled(cfg: ReturnType<typeof buildQaGatewayConfig>) {
+  expect(cfg.plugins?.allow).toContain("qa-lab");
+  expect(cfg.plugins?.entries?.["qa-lab"]).toEqual({ enabled: true });
+}
+
 describe("buildQaGatewayConfig", () => {
   it("keeps mock-openai as the default provider lane", () => {
     const cfg = buildQaGatewayConfig({
@@ -78,7 +83,8 @@ describe("buildQaGatewayConfig", () => {
     expect(cfg.models?.providers?.openai?.request).toEqual({ allowPrivateNetwork: true });
     expect(cfg.models?.providers?.anthropic?.baseUrl).toBe("http://127.0.0.1:44080");
     expect(cfg.models?.providers?.anthropic?.request).toEqual({ allowPrivateNetwork: true });
-    expect(cfg.plugins?.allow).toEqual(["acpx", "memory-core", "qa-channel"]);
+    expect(cfg.plugins?.allow).toEqual(["acpx", "memory-core", "qa-lab", "qa-channel"]);
+    expectQaLabPluginEnabled(cfg);
     expect(cfg.plugins?.slots?.memory).toBe("memory-core");
     expect(cfg.plugins?.entries?.acpx).toEqual({
       enabled: true,
@@ -124,7 +130,7 @@ describe("buildQaGatewayConfig", () => {
     expect(cfg.models?.providers?.anthropic?.models.map((model) => model.id)).toContain(
       "claude-opus-4-8",
     );
-    expect(cfg.plugins?.allow).toEqual(["acpx", "memory-core"]);
+    expect(cfg.plugins?.allow).toEqual(["acpx", "memory-core", "qa-lab"]);
   });
 
   it("falls back to provider defaults for blank model refs", () => {
@@ -175,7 +181,7 @@ describe("buildQaGatewayConfig", () => {
       transportConfig: {},
     });
 
-    expect(cfg.plugins?.allow).toEqual(["acpx", "memory-core"]);
+    expect(cfg.plugins?.allow).toEqual(["acpx", "memory-core", "qa-lab"]);
     expect(cfg.plugins?.entries?.["qa-channel"]).toBeUndefined();
     expect(cfg.channels?.["qa-channel"]).toBeUndefined();
   });
@@ -191,7 +197,13 @@ describe("buildQaGatewayConfig", () => {
       ...createQaChannelTransportParams(),
     });
 
-    expect(cfg.plugins?.allow).toEqual(["acpx", "memory-core", "active-memory", "qa-channel"]);
+    expect(cfg.plugins?.allow).toEqual([
+      "acpx",
+      "memory-core",
+      "qa-lab",
+      "active-memory",
+      "qa-channel",
+    ]);
     expect(cfg.plugins?.entries?.["active-memory"]).toEqual({ enabled: true });
   });
 
@@ -213,7 +225,7 @@ describe("buildQaGatewayConfig", () => {
     expect(getModelFallbacks(cfg.agents?.defaults?.model)).toBeUndefined();
     expect(getModelFallbacks(cfg.agents?.list?.[0]?.model)).toBeUndefined();
     expect(cfg.models).toBeUndefined();
-    expect(cfg.plugins?.allow).toEqual(["acpx", "memory-core", "openai", "qa-channel"]);
+    expect(cfg.plugins?.allow).toEqual(["acpx", "memory-core", "qa-lab", "openai", "qa-channel"]);
     expect(cfg.plugins?.entries?.openai).toEqual({ enabled: true });
     expect(cfg.agents?.defaults?.models?.["openai/gpt-5.5"]).toEqual({
       params: { transport: "sse", openaiWsWarmup: false, fastMode: true },
@@ -236,6 +248,7 @@ describe("buildQaGatewayConfig", () => {
     expect(cfg.plugins?.allow).toEqual([
       "acpx",
       "memory-core",
+      "qa-lab",
       "anthropic",
       "google",
       "qa-channel",
@@ -261,7 +274,7 @@ describe("buildQaGatewayConfig", () => {
     });
 
     expect(getPrimaryModel(cfg.agents?.defaults?.model)).toBe("codex-cli/test-model");
-    expect(cfg.plugins?.allow).toEqual(["acpx", "memory-core", "openai", "qa-channel"]);
+    expect(cfg.plugins?.allow).toEqual(["acpx", "memory-core", "qa-lab", "openai", "qa-channel"]);
     expect(cfg.plugins?.entries?.openai).toEqual({ enabled: true });
     expect(cfg.plugins?.entries?.["codex-cli"]).toBeUndefined();
   });
@@ -301,7 +314,7 @@ describe("buildQaGatewayConfig", () => {
 
     expect(cfg.models?.mode).toBe("merge");
     expect(cfg.models?.providers?.["custom-openai"]?.api).toBe("openai-responses");
-    expect(cfg.plugins?.allow).toEqual(["acpx", "memory-core", "openai", "qa-channel"]);
+    expect(cfg.plugins?.allow).toEqual(["acpx", "memory-core", "qa-lab", "openai", "qa-channel"]);
   });
 
   it("can set a QA default thinking level for judge turns", () => {

extensions/qa-lab/src/qa-gateway-config.ts

@@ -23,6 +23,7 @@ export const DEFAULT_QA_CONTROL_UI_ALLOWED_ORIGINS = Object.freeze([
 ]);
 
 export const QA_BASE_RUNTIME_PLUGIN_IDS = Object.freeze(["acpx", "memory-core"]);
+export const QA_LAB_PLUGIN_ID = "qa-lab";
 
 export function mergeQaControlUiAllowedOrigins(extraOrigins?: string[]) {
   const normalizedExtra = (extraOrigins ?? [])
@@ -112,7 +113,12 @@ export function buildQaGatewayConfig(params: {
     transportPluginIds.map((pluginId) => [pluginId, { enabled: true }]),
   );
   const allowedPlugins = [
-    ...new Set([...QA_BASE_RUNTIME_PLUGIN_IDS, ...selectedPluginIds, ...transportPluginIds]),
+    ...new Set([
+      ...QA_BASE_RUNTIME_PLUGIN_IDS,
+      QA_LAB_PLUGIN_ID,
+      ...selectedPluginIds,
+      ...transportPluginIds,
+    ]),
   ];
   const resolveModelParams = (modelRef: string) =>
     provider.resolveModelParams({
@@ -143,6 +149,9 @@ export function buildQaGatewayConfig(params: {
         "memory-core": {
           enabled: true,
         },
+        [QA_LAB_PLUGIN_ID]: {
+          enabled: true,
+        },
         ...pluginEntries,
         ...transportPluginEntries,
       },

extensions/qa-lab/src/scorecard-taxonomy.ts

@@ -0,0 +1,737 @@
+// Qa Lab plugin module validates the scorecard evidence mapping overlay.
+import fs from "node:fs";
+import path from "node:path";
+import YAML from "yaml";
+import { z } from "zod";
+import type { QaSeedScenarioWithSource } from "./scenario-catalog.js";
+
+export const QA_SCORECARD_TAXONOMY_PATH = "taxonomy-mappings.yaml";
+export const QA_MATURITY_TAXONOMY_PATH = "taxonomy.yaml";
+
+const qaScorecardIdSchema = z
+  .string()
+  .trim()
+  .regex(/^[a-z0-9]+(?:[.-][a-z0-9]+)*$/, {
+    message: "scorecard and coverage ids must use lowercase dotted or dashed tokens",
+  });
+
+function isRepoRootRelativeRef(value: string) {
+  return !path.isAbsolute(value) && value.split(/[\\/]+/u).every((part) => part !== "..");
+}
+
+const qaScorecardRepoRefSchema = z
+  .string()
+  .trim()
+  .min(1)
+  .regex(/^[A-Za-z0-9._/-]+$/, {
+    message: "repo refs must be repo-root relative paths",
+  })
+  .refine(isRepoRootRelativeRef, {
+    message: "repo refs must not be absolute or contain parent-directory segments",
+  });
+
+const qaScorecardFreshnessRuleSchema = z.enum([
+  "target-ref",
+  "target-ref-and-release-package",
+  "release-candidate",
+  "latest-advisory-run",
+]);
+
+const qaScorecardSupportStatusSchema = z.enum(["lts-included", "deferred", "advisory"]);
+
+const qaScorecardTaxonomyRefSchema = z
+  .object({
+    sourcePath: qaScorecardRepoRefSchema,
+    version: z.number().int().positive(),
+    processVersion: z.number().int().positive(),
+    snapshotDate: z.string().trim().min(1),
+    sourceRef: z.string().trim().min(1),
+  })
+  .strict();
+
+const qaScorecardProfileSchema = z.object({
+  id: qaScorecardIdSchema,
+  description: z.string().trim().min(1),
+  categoryIds: z.array(qaScorecardIdSchema).default([]),
+});
+
+const qaScorecardCategorySchema = z.object({
+  id: qaScorecardIdSchema,
+  taxonomySurfaceId: qaScorecardIdSchema,
+  taxonomyCategoryName: z.string().trim().min(1),
+  supportStatus: qaScorecardSupportStatusSchema,
+  releaseBlocking: z.boolean(),
+  requirement: z.string().trim().min(1),
+  evidenceRequired: z.string().trim().min(1),
+  evidence: z.object({
+    profiles: z.array(qaScorecardIdSchema).default([]),
+    liveProofRequired: z.boolean(),
+    freshness: qaScorecardFreshnessRuleSchema,
+    coverageIds: z.array(qaScorecardIdSchema).default([]),
+    scenarioRefs: z.array(qaScorecardRepoRefSchema).default([]),
+    docsRefs: z.array(qaScorecardRepoRefSchema).default([]),
+    codeRefs: z.array(qaScorecardRepoRefSchema).default([]),
+    notes: z.string().trim().min(1).optional(),
+  }),
+});
+
+const qaScorecardTaxonomySchema = z
+  .object({
+    version: z.literal(1),
+    id: qaScorecardIdSchema,
+    title: z.string().trim().min(1),
+    taxonomy: qaScorecardTaxonomyRefSchema,
+    scoreSnapshotRef: qaScorecardRepoRefSchema.optional(),
+    status: z.enum(["initial", "candidate", "active"]),
+    notes: z.string().trim().min(1).optional(),
+    profiles: z.array(qaScorecardProfileSchema).min(1),
+    categories: z.array(qaScorecardCategorySchema).min(1),
+  })
+  .superRefine((taxonomy, ctx) => {
+    const seenProfileIds = new Set<string>();
+    for (const [profileIndex, profile] of taxonomy.profiles.entries()) {
+      if (seenProfileIds.has(profile.id)) {
+        ctx.addIssue({
+          code: z.ZodIssueCode.custom,
+          path: ["profiles", profileIndex, "id"],
+          message: `duplicate scorecard profile id: ${profile.id}`,
+        });
+      }
+      seenProfileIds.add(profile.id);
+
+      const seenProfileCategoryIds = new Set<string>();
+      for (const [categoryIndex, categoryId] of profile.categoryIds.entries()) {
+        if (seenProfileCategoryIds.has(categoryId)) {
+          ctx.addIssue({
+            code: z.ZodIssueCode.custom,
+            path: ["profiles", profileIndex, "categoryIds", categoryIndex],
+            message: `duplicate category id in profile ${profile.id}: ${categoryId}`,
+          });
+        }
+        seenProfileCategoryIds.add(categoryId);
+      }
+    }
+
+    const seenCategoryIds = new Set<string>();
+    for (const [categoryIndex, category] of taxonomy.categories.entries()) {
+      if (seenCategoryIds.has(category.id)) {
+        ctx.addIssue({
+          code: z.ZodIssueCode.custom,
+          path: ["categories", categoryIndex, "id"],
+          message: `duplicate scorecard category id: ${category.id}`,
+        });
+      }
+      seenCategoryIds.add(category.id);
+
+      if (category.supportStatus === "lts-included" && !category.releaseBlocking) {
+        ctx.addIssue({
+          code: z.ZodIssueCode.custom,
+          path: ["categories", categoryIndex, "releaseBlocking"],
+          message: `LTS-included category ${category.id} must be release-blocking`,
+        });
+      }
+      if (category.supportStatus !== "lts-included" && category.releaseBlocking) {
+        ctx.addIssue({
+          code: z.ZodIssueCode.custom,
+          path: ["categories", categoryIndex, "releaseBlocking"],
+          message: `${category.supportStatus} category ${category.id} must not be release-blocking`,
+        });
+      }
+
+      const seenCoverageIds = new Set<string>();
+      for (const [coverageIndex, coverageId] of category.evidence.coverageIds.entries()) {
+        if (seenCoverageIds.has(coverageId)) {
+          ctx.addIssue({
+            code: z.ZodIssueCode.custom,
+            path: ["categories", categoryIndex, "evidence", "coverageIds", coverageIndex],
+            message: `duplicate coverage id in category ${category.id}: ${coverageId}`,
+          });
+        }
+        seenCoverageIds.add(coverageId);
+      }
+    }
+  });
+
+const qaMaturityCategorySchema = z.object({
+  name: z.string().trim().min(1),
+});
+
+const qaMaturitySurfaceSchema = z.object({
+  id: qaScorecardIdSchema,
+  name: z.string().trim().min(1),
+  level: z.string().trim().min(1).optional(),
+  level_code: z.string().trim().min(1).optional(),
+  categories: z.array(qaMaturityCategorySchema).default([]),
+});
+
+const qaMaturityTaxonomySchema = z.object({
+  version: z.number(),
+  title: z.string().trim().min(1),
+  surfaces: z.array(qaMaturitySurfaceSchema).default([]),
+});
+
+export type QaScorecardTaxonomy = z.infer<typeof qaScorecardTaxonomySchema>;
+export type QaScorecardTaxonomyCategory = QaScorecardTaxonomy["categories"][number];
+type QaMaturityTaxonomy = z.infer<typeof qaMaturityTaxonomySchema>;
+
+export type QaScorecardValidationIssueCode =
+  | "coverage-id-not-found"
+  | "scenario-ref-not-found"
+  | "scenario-ref-not-covered-by-category"
+  | "docs-ref-not-found"
+  | "code-ref-not-found"
+  | "taxonomy-ref-not-found"
+  | "taxonomy-category-ref-not-found"
+  | "profile-category-ref-not-found"
+  | "score-snapshot-ref-not-found"
+  | "blocking-category-without-evidence-mapping"
+  | "non-advisory-category-missing-profile-membership"
+  | "release-blocking-category-missing-release-profile"
+  | "advisory-category-has-profile-membership"
+  | "profile-ref-not-found"
+  | "category-profile-missing-top-level-membership"
+  | "profile-membership-missing-category-profile"
+  | "taxonomy-fixture-not-found";
+
+export type QaScorecardValidationIssue = {
+  code: QaScorecardValidationIssueCode;
+  severity: "warning";
+  categoryId?: string;
+  ref?: string;
+  message: string;
+};
+
+export type QaScorecardCategoryMappingReport = {
+  id: string;
+  taxonomySurfaceId: string;
+  taxonomyCategoryName: string;
+  supportStatus: string;
+  releaseBlocking: boolean;
+  mappingStatus: "mapped" | "partial" | "missing";
+  profiles: string[];
+  liveProofRequired: boolean;
+  freshness: string;
+  coverageIds: string[];
+  scenarioRefs: string[];
+  missingCoverageIds: string[];
+  missingScenarioRefs: string[];
+};
+
+export type QaScorecardProfileReport = {
+  id: string;
+  categoryIds: string[];
+};
+
+export type QaScorecardTaxonomyReport = {
+  taxonomyPath: string | null;
+  taxonomyId: string | null;
+  title: string | null;
+  taxonomy: {
+    sourcePath: string;
+    version: number;
+    processVersion: number;
+    snapshotDate: string;
+    sourceRef: string;
+  } | null;
+  scoreSnapshotRef: string | null;
+  status: string | null;
+  profileCount: number;
+  profiles: QaScorecardProfileReport[];
+  categoryCount: number;
+  releaseBlockingCategoryCount: number;
+  advisoryCategoryCount: number;
+  ltsIncludedCategoryCount: number;
+  deferredCategoryCount: number;
+  mappedCoverageIdCount: number;
+  mappedScenarioCount: number;
+  unmappedCoverageIdCount: number;
+  unmappedCoverageIds: string[];
+  validationIssueCount: number;
+  validationIssues: QaScorecardValidationIssue[];
+  categories: QaScorecardCategoryMappingReport[];
+};
+
+function walkUpDirectories(start: string): string[] {
+  const roots: string[] = [];
+  let current = path.resolve(start);
+  while (true) {
+    roots.push(current);
+    const parent = path.dirname(current);
+    if (parent === current) {
+      return roots;
+    }
+    current = parent;
+  }
+}
+
+function resolveRepoPath(relativePath: string, kind: "file" | "directory" = "file") {
+  for (const dir of walkUpDirectories(import.meta.dirname)) {
+    const candidate = path.join(dir, relativePath);
+    if (!fs.existsSync(candidate)) {
+      continue;
+    }
+    const stat = fs.statSync(candidate);
+    if ((kind === "file" && stat.isFile()) || (kind === "directory" && stat.isDirectory())) {
+      return candidate;
+    }
+  }
+  return null;
+}
+
+function repoRootFromMappingPath(mappingPath: string) {
+  return path.dirname(mappingPath);
+}
+
+function formatZodIssuePath(pathLocal: PropertyKey[]) {
+  return pathLocal.length ? pathLocal.map(String).join(".") : "<root>";
+}
+
+export function parseQaScorecardTaxonomy(value: unknown, label = QA_SCORECARD_TAXONOMY_PATH) {
+  const parsed = qaScorecardTaxonomySchema.safeParse(value);
+  if (parsed.success) {
+    return parsed.data;
+  }
+  const issues = parsed.error.issues
+    .map((issue) => `${formatZodIssuePath(issue.path)}: ${issue.message}`)
+    .join("; ");
+  throw new Error(`${label}: ${issues}`);
+}
+
+export function readQaScorecardTaxonomy(): QaScorecardTaxonomy | null {
+  const taxonomyPath = resolveRepoPath(QA_SCORECARD_TAXONOMY_PATH, "file");
+  if (!taxonomyPath) {
+    return null;
+  }
+  return parseQaScorecardTaxonomy(
+    YAML.parse(fs.readFileSync(taxonomyPath, "utf8")) as unknown,
+    QA_SCORECARD_TAXONOMY_PATH,
+  );
+}
+
+function parseQaMaturityTaxonomy(value: unknown, label = QA_MATURITY_TAXONOMY_PATH) {
+  const parsed = qaMaturityTaxonomySchema.safeParse(value);
+  if (parsed.success) {
+    return parsed.data;
+  }
+  const issues = parsed.error.issues
+    .map((issue) => `${formatZodIssuePath(issue.path)}: ${issue.message}`)
+    .join("; ");
+  throw new Error(`${label}: ${issues}`);
+}
+
+function readQaMaturityTaxonomy(repoRoot: string | undefined, taxonomySourcePath: string) {
+  const taxonomyPath = repoRoot
+    ? path.join(repoRoot, taxonomySourcePath)
+    : resolveRepoPath(taxonomySourcePath);
+  if (!taxonomyPath || !fs.existsSync(taxonomyPath)) {
+    return null;
+  }
+  return parseQaMaturityTaxonomy(
+    YAML.parse(fs.readFileSync(taxonomyPath, "utf8")) as unknown,
+    taxonomySourcePath,
+  );
+}
+
+function maturityCategoryKey(surfaceId: string, categoryName: string) {
+  return `${surfaceId}\0${categoryName}`;
+}
+
+function buildMaturityCategoryKeys(taxonomy: QaMaturityTaxonomy | null) {
+  const categoryKeys = new Set<string>();
+  if (!taxonomy) {
+    return categoryKeys;
+  }
+  for (const surface of taxonomy.surfaces) {
+    for (const category of surface.categories) {
+      categoryKeys.add(maturityCategoryKey(surface.id, category.name));
+    }
+  }
+  return categoryKeys;
+}
+
+function scenarioCoverageIds(scenario: QaSeedScenarioWithSource) {
+  return [...(scenario.coverage?.primary ?? []), ...(scenario.coverage?.secondary ?? [])];
+}
+
+function pathExists(repoRoot: string | undefined, relativePath: string) {
+  if (!isRepoRootRelativeRef(relativePath)) {
+    return false;
+  }
+  return repoRoot ? fs.existsSync(path.join(repoRoot, relativePath)) : true;
+}
+
+function reportMissingRepoRefs(params: {
+  repoRoot: string | undefined;
+  categoryId: string;
+  refs: readonly string[];
+  code: "docs-ref-not-found" | "code-ref-not-found";
+  label: "docs" | "code";
+  issues: QaScorecardValidationIssue[];
+}) {
+  for (const ref of params.refs) {
+    if (pathExists(params.repoRoot, ref)) {
+      continue;
+    }
+    params.issues.push({
+      code: params.code,
+      severity: "warning",
+      categoryId: params.categoryId,
+      ref,
+      message: `${params.categoryId} references missing ${params.label} ref ${ref}`,
+    });
+  }
+}
+
+export function buildQaScorecardTaxonomyReport(params: {
+  taxonomy: QaScorecardTaxonomy | null;
+  taxonomyPath?: string | null;
+  repoRoot?: string;
+  scenarios: readonly QaSeedScenarioWithSource[];
+}): QaScorecardTaxonomyReport {
+  if (!params.taxonomy) {
+    const issue = {
+      code: "taxonomy-fixture-not-found",
+      severity: "warning",
+      ref: QA_SCORECARD_TAXONOMY_PATH,
+      message: `Scorecard evidence mapping not found at ${QA_SCORECARD_TAXONOMY_PATH}`,
+    } satisfies QaScorecardValidationIssue;
+    return {
+      taxonomyPath: params.taxonomyPath ?? null,
+      taxonomyId: null,
+      title: null,
+      taxonomy: null,
+      scoreSnapshotRef: null,
+      status: null,
+      profileCount: 0,
+      profiles: [],
+      categoryCount: 0,
+      releaseBlockingCategoryCount: 0,
+      advisoryCategoryCount: 0,
+      ltsIncludedCategoryCount: 0,
+      deferredCategoryCount: 0,
+      mappedCoverageIdCount: 0,
+      mappedScenarioCount: 0,
+      unmappedCoverageIdCount: 0,
+      unmappedCoverageIds: [],
+      validationIssueCount: 1,
+      validationIssues: [issue],
+      categories: [],
+    };
+  }
+
+  const coverageIdsByScenarioRef = new Map(
+    params.scenarios.map((scenario) => [
+      scenario.sourcePath,
+      new Set(scenarioCoverageIds(scenario)),
+    ]),
+  );
+  const scenarioRefsByCoverageId = new Map<string, Set<string>>();
+  for (const scenario of params.scenarios) {
+    for (const coverageId of scenarioCoverageIds(scenario)) {
+      const refs = scenarioRefsByCoverageId.get(coverageId) ?? new Set<string>();
+      refs.add(scenario.sourcePath);
+      scenarioRefsByCoverageId.set(coverageId, refs);
+    }
+  }
+
+  const issues: QaScorecardValidationIssue[] = [];
+  const categories: QaScorecardCategoryMappingReport[] = [];
+  const mappedCoverageIds = new Set<string>();
+  const mappedScenarioRefs = new Set<string>();
+  const categoryIds = new Set(params.taxonomy.categories.map((category) => category.id));
+  const profileIds = new Set(params.taxonomy.profiles.map((profile) => profile.id));
+  const maturityTaxonomy = readQaMaturityTaxonomy(
+    params.repoRoot,
+    params.taxonomy.taxonomy.sourcePath,
+  );
+  const maturityCategoryKeys = buildMaturityCategoryKeys(maturityTaxonomy);
+  const profileCategoryIdsByCategoryId = new Map<string, Set<string>>();
+  const profiles = params.taxonomy.profiles.map((profile) => {
+    for (const categoryId of profile.categoryIds) {
+      if (!categoryIds.has(categoryId)) {
+        issues.push({
+          code: "profile-category-ref-not-found",
+          severity: "warning",
+          ref: categoryId,
+          message: `${profile.id} profile references missing executable scorecard category ${categoryId}`,
+        });
+        continue;
+      }
+      const categoryProfileIds =
+        profileCategoryIdsByCategoryId.get(categoryId) ?? new Set<string>();
+      categoryProfileIds.add(profile.id);
+      profileCategoryIdsByCategoryId.set(categoryId, categoryProfileIds);
+    }
+
+    return {
+      id: profile.id,
+      categoryIds: profile.categoryIds.filter((categoryId) => categoryIds.has(categoryId)),
+    };
+  });
+
+  if (!pathExists(params.repoRoot, params.taxonomy.taxonomy.sourcePath) || !maturityTaxonomy) {
+    issues.push({
+      code: "taxonomy-ref-not-found",
+      severity: "warning",
+      ref: params.taxonomy.taxonomy.sourcePath,
+      message: `Scorecard executable mapping references missing maturity taxonomy ${params.taxonomy.taxonomy.sourcePath}`,
+    });
+  }
+  if (
+    params.taxonomy.scoreSnapshotRef &&
+    !pathExists(params.repoRoot, params.taxonomy.scoreSnapshotRef)
+  ) {
+    issues.push({
+      code: "score-snapshot-ref-not-found",
+      severity: "warning",
+      ref: params.taxonomy.scoreSnapshotRef,
+      message: `Scorecard executable mapping references missing score snapshot ${params.taxonomy.scoreSnapshotRef}`,
+    });
+  }
+
+  for (const category of params.taxonomy.categories) {
+    const missingCoverageIds: string[] = [];
+    const missingScenarioRefs: string[] = [];
+    const declaredProfileIds = new Set(category.evidence.profiles);
+    const declaredKnownProfileIds = new Set(
+      [...declaredProfileIds].filter((profileId) => profileIds.has(profileId)),
+    );
+    const membershipProfileIds =
+      profileCategoryIdsByCategoryId.get(category.id) ?? new Set<string>();
+    const sortedMembershipProfileIds = [...membershipProfileIds].toSorted();
+    const maturityKey = maturityCategoryKey(
+      category.taxonomySurfaceId,
+      category.taxonomyCategoryName,
+    );
+
+    if (maturityTaxonomy && !maturityCategoryKeys.has(maturityKey)) {
+      issues.push({
+        code: "taxonomy-category-ref-not-found",
+        severity: "warning",
+        categoryId: category.id,
+        ref: `${category.taxonomySurfaceId}/${category.taxonomyCategoryName}`,
+        message: `${category.id} references missing maturity taxonomy category ${category.taxonomySurfaceId}/${category.taxonomyCategoryName}`,
+      });
+    }
+
+    for (const profileId of declaredProfileIds) {
+      if (!profileIds.has(profileId)) {
+        issues.push({
+          code: "profile-ref-not-found",
+          severity: "warning",
+          categoryId: category.id,
+          ref: profileId,
+          message: `${category.id} declares profile ${profileId}, but taxonomy-mappings.yaml has no matching top-level profile`,
+        });
+        continue;
+      }
+
+      if (!membershipProfileIds.has(profileId)) {
+        issues.push({
+          code: "category-profile-missing-top-level-membership",
+          severity: "warning",
+          categoryId: category.id,
+          ref: profileId,
+          message: `${category.id} declares ${profileId} evidence, but the taxonomy profile does not include the category`,
+        });
+      }
+    }
+
+    for (const profileId of membershipProfileIds) {
+      if (!declaredProfileIds.has(profileId)) {
+        issues.push({
+          code: "profile-membership-missing-category-profile",
+          severity: "warning",
+          categoryId: category.id,
+          ref: profileId,
+          message: `${category.id} belongs to the ${profileId} taxonomy profile, but its evidence profiles do not declare that selector`,
+        });
+      }
+    }
+
+    if (category.releaseBlocking && !membershipProfileIds.has("release")) {
+      issues.push({
+        code: "release-blocking-category-missing-release-profile",
+        severity: "warning",
+        categoryId: category.id,
+        ref: "release",
+        message: `${category.id} is release-blocking but is not selected by the release profile`,
+      });
+    }
+
+    if (
+      category.supportStatus === "advisory" &&
+      (membershipProfileIds.size > 0 || declaredProfileIds.size > 0)
+    ) {
+      const runnableProfiles = [
+        ...new Set([...membershipProfileIds, ...declaredProfileIds]),
+      ].toSorted();
+      issues.push({
+        code: "advisory-category-has-profile-membership",
+        severity: "warning",
+        categoryId: category.id,
+        message: `${category.id} is advisory metadata but belongs to runnable profile(s): ${runnableProfiles.join(", ")}`,
+      });
+    }
+
+    if (
+      category.supportStatus !== "advisory" &&
+      membershipProfileIds.size === 0 &&
+      declaredKnownProfileIds.size === 0
+    ) {
+      issues.push({
+        code: "non-advisory-category-missing-profile-membership",
+        severity: "warning",
+        categoryId: category.id,
+        message: `${category.id} is ${category.supportStatus} but has no runnable profile membership`,
+      });
+    }
+
+    for (const coverageId of category.evidence.coverageIds) {
+      const scenarioRefs = scenarioRefsByCoverageId.get(coverageId);
+      if (!scenarioRefs) {
+        missingCoverageIds.push(coverageId);
+        issues.push({
+          code: "coverage-id-not-found",
+          severity: "warning",
+          categoryId: category.id,
+          ref: coverageId,
+          message: `${category.id} maps missing coverage id ${coverageId}`,
+        });
+        continue;
+      }
+      mappedCoverageIds.add(coverageId);
+      for (const scenarioRef of scenarioRefs) {
+        mappedScenarioRefs.add(scenarioRef);
+      }
+    }
+
+    const categoryCoverageIds = new Set(category.evidence.coverageIds);
+    for (const scenarioRef of category.evidence.scenarioRefs) {
+      const scenarioCoverage = coverageIdsByScenarioRef.get(scenarioRef);
+      if (!scenarioCoverage) {
+        missingScenarioRefs.push(scenarioRef);
+        issues.push({
+          code: "scenario-ref-not-found",
+          severity: "warning",
+          categoryId: category.id,
+          ref: scenarioRef,
+          message: `${category.id} references missing scenario ${scenarioRef}`,
+        });
+        continue;
+      }
+      mappedScenarioRefs.add(scenarioRef);
+      if (
+        categoryCoverageIds.size > 0 &&
+        ![...scenarioCoverage].some((coverageId) => categoryCoverageIds.has(coverageId))
+      ) {
+        issues.push({
+          code: "scenario-ref-not-covered-by-category",
+          severity: "warning",
+          categoryId: category.id,
+          ref: scenarioRef,
+          message: `${category.id} references ${scenarioRef} without one of the category coverage IDs`,
+        });
+      }
+    }
+
+    reportMissingRepoRefs({
+      repoRoot: params.repoRoot,
+      categoryId: category.id,
+      refs: category.evidence.docsRefs,
+      code: "docs-ref-not-found",
+      label: "docs",
+      issues,
+    });
+    reportMissingRepoRefs({
+      repoRoot: params.repoRoot,
+      categoryId: category.id,
+      refs: category.evidence.codeRefs,
+      code: "code-ref-not-found",
+      label: "code",
+      issues,
+    });
+
+    if (
+      category.releaseBlocking &&
+      category.evidence.coverageIds.length === 0 &&
+      category.evidence.scenarioRefs.length === 0
+    ) {
+      issues.push({
+        code: "blocking-category-without-evidence-mapping",
+        severity: "warning",
+        categoryId: category.id,
+        message: `${category.id} is release-blocking but has no coverage IDs or scenario refs`,
+      });
+    }
+
+    const mappingStatus =
+      category.evidence.coverageIds.length === 0 && category.evidence.scenarioRefs.length === 0
+        ? "missing"
+        : missingCoverageIds.length > 0 || missingScenarioRefs.length > 0
+          ? "partial"
+          : "mapped";
+    categories.push({
+      id: category.id,
+      taxonomySurfaceId: category.taxonomySurfaceId,
+      taxonomyCategoryName: category.taxonomyCategoryName,
+      supportStatus: category.supportStatus,
+      releaseBlocking: category.releaseBlocking,
+      mappingStatus,
+      profiles: sortedMembershipProfileIds,
+      liveProofRequired: category.evidence.liveProofRequired,
+      freshness: category.evidence.freshness,
+      coverageIds: [...category.evidence.coverageIds],
+      scenarioRefs: [...category.evidence.scenarioRefs],
+      missingCoverageIds,
+      missingScenarioRefs,
+    });
+  }
+
+  const allCoverageIds = [...scenarioRefsByCoverageId.keys()].toSorted();
+  const unmappedCoverageIds = allCoverageIds.filter(
+    (coverageId) => !mappedCoverageIds.has(coverageId),
+  );
+
+  return {
+    taxonomyPath: params.taxonomyPath ?? QA_SCORECARD_TAXONOMY_PATH,
+    taxonomyId: params.taxonomy.id,
+    title: params.taxonomy.title,
+    taxonomy: params.taxonomy.taxonomy,
+    scoreSnapshotRef: params.taxonomy.scoreSnapshotRef ?? null,
+    status: params.taxonomy.status,
+    profileCount: params.taxonomy.profiles.length,
+    profiles,
+    categoryCount: params.taxonomy.categories.length,
+    releaseBlockingCategoryCount: params.taxonomy.categories.filter(
+      (category) => category.releaseBlocking,
+    ).length,
+    advisoryCategoryCount: params.taxonomy.categories.filter(
+      (category) => category.supportStatus === "advisory",
+    ).length,
+    ltsIncludedCategoryCount: params.taxonomy.categories.filter(
+      (category) => category.supportStatus === "lts-included",
+    ).length,
+    deferredCategoryCount: params.taxonomy.categories.filter(
+      (category) => category.supportStatus === "deferred",
+    ).length,
+    mappedCoverageIdCount: mappedCoverageIds.size,
+    mappedScenarioCount: mappedScenarioRefs.size,
+    unmappedCoverageIdCount: unmappedCoverageIds.length,
+    unmappedCoverageIds,
+    validationIssueCount: issues.length,
+    validationIssues: issues,
+    categories: categories.toSorted((left, right) => left.id.localeCompare(right.id)),
+  };
+}
+
+export function readQaScorecardTaxonomyReport(scenarios: readonly QaSeedScenarioWithSource[]) {
+  const taxonomyPath = resolveRepoPath(QA_SCORECARD_TAXONOMY_PATH, "file");
+  const taxonomy = readQaScorecardTaxonomy();
+  return buildQaScorecardTaxonomyReport({
+    taxonomy,
+    taxonomyPath: taxonomyPath ? QA_SCORECARD_TAXONOMY_PATH : null,
+    repoRoot: taxonomyPath ? repoRootFromMappingPath(taxonomyPath) : undefined,
+    scenarios,
+  });
+}

extensions/qa-lab/src/suite-runtime-agent-process.ts

@@ -2,6 +2,7 @@
 import { spawn } from "node:child_process";
 import { randomUUID } from "node:crypto";
 import path from "node:path";
+import { formatErrorMessage } from "openclaw/plugin-sdk/error-runtime";
 import { resolveTimerTimeoutMs } from "openclaw/plugin-sdk/number-runtime";
 import {
   appendQaChildOutput,
@@ -12,6 +13,7 @@ import {
   QA_CHILD_STDOUT_MAX_BYTES,
   readQaChildOutput,
 } from "./child-output.js";
+import { QaSuiteInfraError } from "./errors.js";
 import { resolveQaNodeExecPath } from "./node-exec.js";
 import { liveTurnTimeoutMs } from "./suite-runtime-agent-common.js";
 import { waitForGatewayHealthy, waitForTransportReady } from "./suite-runtime-gateway.js";
@@ -103,7 +105,9 @@ async function runQaCli(
     const timeoutMs = resolveTimerTimeoutMs(opts?.timeoutMs, 60_000);
     const timeout = setTimeout(() => {
       child.kill("SIGKILL");
-      reject(new Error(`qa cli timed out: openclaw ${args.join(" ")}`));
+      reject(
+        new QaSuiteInfraError("qa_cli_timeout", `qa cli timed out: openclaw ${args.join(" ")}`),
+      );
     }, timeoutMs);
     child.stdout.on("data", (chunk) => appendQaChildOutput(stdout, chunk));
     child.stderr.on("data", (chunk) => appendQaChildOutputTail(stderr, chunk));
@@ -187,16 +191,24 @@ async function waitForAgentRun(
   runId: string,
   timeoutMs = 30_000,
 ) {
-  return (await env.gateway.call(
-    "agent.wait",
-    {
-      runId,
-      timeoutMs,
-    },
-    {
-      timeoutMs: resolveQaGatewayTimeoutWithGraceMs(timeoutMs),
-    },
-  )) as { status?: string; error?: string };
+  try {
+    return (await env.gateway.call(
+      "agent.wait",
+      {
+        runId,
+        timeoutMs,
+      },
+      {
+        timeoutMs: resolveQaGatewayTimeoutWithGraceMs(timeoutMs),
+      },
+    )) as { status?: string; error?: string };
+  } catch (error) {
+    throw new QaSuiteInfraError(
+      "agent_wait_failed",
+      `agent.wait failed: ${formatErrorMessage(error)}`,
+      { cause: error },
+    );
+  }
 }
 
 async function listCronJobs(env: Pick<QaSuiteRuntimeEnv, "gateway">) {

extensions/qa-lab/src/suite-runtime-gateway.ts

@@ -3,6 +3,7 @@ import { setTimeout as sleep } from "node:timers/promises";
 import { formatErrorMessage } from "openclaw/plugin-sdk/error-runtime";
 import { fetchWithSsrFGuard } from "openclaw/plugin-sdk/ssrf-runtime";
 import { isRecord as isPlainObject } from "openclaw/plugin-sdk/string-coerce-runtime";
+import { QaSuiteInfraError } from "./errors.js";
 import { applyQaMergePatch } from "./suite-merge-patch.js";
 import { liveTurnTimeoutMs } from "./suite-runtime-agent-common.js";
 import type { QaConfigSnapshot, QaSuiteRuntimeEnv } from "./suite-runtime-types.js";
@@ -50,7 +51,7 @@ async function waitForGatewayHealthy(env: Pick<QaSuiteRuntimeEnv, "gateway">, ti
     }
     await sleep(250);
   }
-  throw new Error(`timed out after ${timeoutMs}ms`);
+  throw new QaSuiteInfraError("gateway_ready_timeout", `timed out after ${timeoutMs}ms`);
 }
 
 async function waitForTransportReady(
@@ -94,7 +95,8 @@ async function waitForConfigRestartSettle(
     await sleep(Math.min(250, Math.max(1, deadline - Date.now())));
   }
 
-  throw new Error(
+  throw new QaSuiteInfraError(
+    "transport_ready_timeout",
     `timed out after ${timeoutMs}ms waiting for config restart readiness${
       lastHealthError ? `: ${formatErrorMessage(lastHealthError)}` : ""
     }`,

extensions/qa-lab/src/suite-summary.test.ts

@@ -72,6 +72,19 @@ describe("qa suite summary helpers", () => {
     ).toBe(1);
   });
 
+  it("counts evidence entry results", () => {
+    const summary = {
+      entries: [
+        { result: { status: "pass" } },
+        { result: { status: "fail" } },
+        { result: { status: "skipped" } },
+      ],
+    };
+
+    expect(readQaSuiteFailedScenarioCountFromSummary(summary)).toBe(1);
+    expect(readQaSuiteFailedOrSkippedScenarioCountFromSummary(summary)).toBe(2);
+  });
+
   it("uses the larger blocking signal when skipped counts and scenarios disagree", () => {
     expect(
       readQaSuiteFailedOrSkippedScenarioCountFromSummary({
@@ -138,7 +151,7 @@ describe("qa suite summary helpers", () => {
 
     try {
       await expect(readQaSuiteFailedScenarioCountFromFile(summaryPath)).rejects.toThrow(
-        "did not include counts.failed or scenarios[].status",
+        "did not include counts.failed, scenarios[].status, or entries[].result.status",
       );
     } finally {
       await fs.rm(outputDir, { recursive: true, force: true });

extensions/qa-lab/src/suite-summary.ts

@@ -1,6 +1,8 @@
 // Qa Lab plugin module implements suite summary behavior.
 import fs from "node:fs/promises";
 import { formatErrorMessage } from "openclaw/plugin-sdk/error-runtime";
+import { QaSuiteArtifactError } from "./errors.js";
+import type { QaEvidenceSummaryJson } from "./evidence-summary.js";
 import type { QaProviderMode } from "./model-selection.js";
 import type { RuntimeId, RuntimeParityResult } from "./runtime-parity.js";
 
@@ -40,6 +42,7 @@ export type QaSuiteSummaryJson = {
       bytes: number;
     }>;
   };
+  evidence?: QaEvidenceSummaryJson;
   run: {
     startedAt: string;
     finishedAt: string;
@@ -58,6 +61,11 @@ export type QaSuiteSummaryJson = {
 };
 
 type QaSuiteScenarioStatus = Pick<QaSuiteSummaryScenario, "status">;
+type QaEvidenceEntryStatus = {
+  result?: {
+    status?: unknown;
+  };
+};
 
 function readNonNegativeCount(value: unknown): number | null {
   return typeof value === "number" && Number.isFinite(value)
@@ -101,17 +109,27 @@ export function readQaSuiteFailedScenarioCountFromSummary(summary: unknown): num
     counts?: {
       failed?: unknown;
     };
+    entries?: QaEvidenceEntryStatus[];
     scenarios?: Array<QaSuiteScenarioStatus>;
   };
   const countedFailures = readNonNegativeCount(payload.counts?.failed);
   const scenarioFailures = Array.isArray(payload.scenarios)
     ? countQaSuiteFailedScenarios(payload.scenarios)
     : null;
+  const evidenceFailures = Array.isArray(payload.entries)
+    ? payload.entries.filter((entry) => entry.result?.status === "fail").length
+    : null;
   if (countedFailures !== null && scenarioFailures !== null) {
-    return Math.max(countedFailures, scenarioFailures);
+    return Math.max(countedFailures, scenarioFailures, evidenceFailures ?? 0);
+  }
+  if (countedFailures !== null && evidenceFailures !== null) {
+    return Math.max(countedFailures, evidenceFailures);
   }
   if (scenarioFailures !== null) {
-    return scenarioFailures;
+    return Math.max(scenarioFailures, evidenceFailures ?? 0);
+  }
+  if (evidenceFailures !== null) {
+    return evidenceFailures;
   }
   return countedFailures;
 }
@@ -127,6 +145,7 @@ export function readQaSuiteFailedOrSkippedScenarioCountFromSummary(
       failed?: unknown;
       skipped?: unknown;
     };
+    entries?: QaEvidenceEntryStatus[];
     scenarios?: Array<QaSuiteScenarioStatus>;
   };
   const countedFailures = readNonNegativeCount(payload.counts?.failed);
@@ -138,11 +157,20 @@ export function readQaSuiteFailedOrSkippedScenarioCountFromSummary(
   const scenarioBlocking = Array.isArray(payload.scenarios)
     ? countQaSuiteFailedOrSkippedScenarios(payload.scenarios)
     : null;
+  const evidenceBlocking = Array.isArray(payload.entries)
+    ? payload.entries.filter((entry) => isQaSuiteBlockingStatus(entry.result?.status)).length
+    : null;
   if (countedBlocking !== null && scenarioBlocking !== null) {
-    return Math.max(countedBlocking, scenarioBlocking);
+    return Math.max(countedBlocking, scenarioBlocking, evidenceBlocking ?? 0);
+  }
+  if (countedBlocking !== null && evidenceBlocking !== null) {
+    return Math.max(countedBlocking, evidenceBlocking);
   }
   if (scenarioBlocking !== null) {
-    return scenarioBlocking;
+    return Math.max(scenarioBlocking, evidenceBlocking ?? 0);
+  }
+  if (evidenceBlocking !== null) {
+    return evidenceBlocking;
   }
   return countedBlocking;
 }
@@ -152,7 +180,8 @@ export async function readQaSuiteFailedScenarioCountFromFile(summaryPath: string
   try {
     summaryText = await fs.readFile(summaryPath, "utf8");
   } catch (error) {
-    throw new Error(
+    throw new QaSuiteArtifactError(
+      "summary_read_failed",
       `Could not read QA summary JSON at ${summaryPath}: ${formatErrorMessage(error)}`,
       { cause: error },
     );
@@ -161,7 +190,8 @@ export async function readQaSuiteFailedScenarioCountFromFile(summaryPath: string
   try {
     payload = JSON.parse(summaryText) as unknown;
   } catch (error) {
-    throw new Error(
+    throw new QaSuiteArtifactError(
+      "summary_parse_failed",
       `Could not parse QA summary JSON at ${summaryPath}: ${formatErrorMessage(error)}`,
       { cause: error },
     );
@@ -170,8 +200,9 @@ export async function readQaSuiteFailedScenarioCountFromFile(summaryPath: string
   if (failedScenarioCount !== null) {
     return failedScenarioCount;
   }
-  throw new Error(
-    `QA summary at ${summaryPath} did not include counts.failed or scenarios[].status.`,
+  throw new QaSuiteArtifactError(
+    "summary_failure_count_missing",
+    `QA summary at ${summaryPath} did not include counts.failed, scenarios[].status, or entries[].result.status.`,
   );
 }
 
@@ -182,7 +213,8 @@ export async function readQaSuiteFailedOrSkippedScenarioCountFromFile(
   try {
     summaryText = await fs.readFile(summaryPath, "utf8");
   } catch (error) {
-    throw new Error(
+    throw new QaSuiteArtifactError(
+      "summary_read_failed",
       `Could not read QA summary JSON at ${summaryPath}: ${formatErrorMessage(error)}`,
       { cause: error },
     );
@@ -191,7 +223,8 @@ export async function readQaSuiteFailedOrSkippedScenarioCountFromFile(
   try {
     payload = JSON.parse(summaryText) as unknown;
   } catch (error) {
-    throw new Error(
+    throw new QaSuiteArtifactError(
+      "summary_parse_failed",
       `Could not parse QA summary JSON at ${summaryPath}: ${formatErrorMessage(error)}`,
       { cause: error },
     );
@@ -200,7 +233,8 @@ export async function readQaSuiteFailedOrSkippedScenarioCountFromFile(
   if (blockingScenarioCount !== null) {
     return blockingScenarioCount;
   }
-  throw new Error(
-    `QA summary at ${summaryPath} did not include counts.failed, counts.skipped, or scenarios[].status.`,
+  throw new QaSuiteArtifactError(
+    "summary_blocking_count_missing",
+    `QA summary at ${summaryPath} did not include counts.failed, counts.skipped, scenarios[].status, or entries[].result.status.`,
   );
 }

extensions/qa-lab/src/suite.summary-json.test.ts

@@ -1,5 +1,6 @@
 // Qa Lab tests cover suite.summary json plugin behavior.
 import { describe, expect, it } from "vitest";
+import { buildQaSuiteEvidenceSummary } from "./evidence-summary.js";
 import { buildQaSuiteSummaryJson } from "./suite.js";
 
 describe("buildQaSuiteSummaryJson", () => {
@@ -103,6 +104,34 @@ describe("buildQaSuiteSummaryJson", () => {
     });
   });
 
+  it("preserves the evidence summary when provided", () => {
+    const evidence = buildQaSuiteEvidenceSummary({
+      artifactPaths: [{ kind: "summary", path: "qa-suite-summary.json" }],
+      scenarioDefinitions: [
+        {
+          id: "dm-chat-baseline",
+          title: "DM baseline conversation",
+          sourcePath: "qa/scenarios/channels/dm-chat-baseline.md",
+          surface: "dm",
+          coverage: {
+            primary: ["channels.dm"],
+          },
+        },
+      ],
+      channelId: "qa-channel",
+      generatedAt: "2026-04-11T00:05:00.000Z",
+      primaryModel: "mock-openai/gpt-5.5",
+      providerMode: "mock-openai",
+      scenarioResults: [{ name: "DM baseline conversation", status: "pass" }],
+    });
+    const json = buildQaSuiteSummaryJson({
+      ...baseParams,
+      evidence,
+    });
+
+    expect(json.evidence).toEqual(evidence);
+  });
+
   it("preserves scenario-level runtime parity payloads", () => {
     const json = buildQaSuiteSummaryJson({
       ...baseParams,

extensions/qa-lab/src/suite.ts

@@ -12,6 +12,7 @@ import {
   type QaReportScenario,
 } from "openclaw/plugin-sdk/qa-runtime";
 import { fetchWithSsrFGuard } from "openclaw/plugin-sdk/ssrf-runtime";
+import { buildQaSuiteEvidenceSummary } from "./evidence-summary.js";
 import { startQaGatewayChild, type QaCliBackendAuthMode } from "./gateway-child.js";
 import type {
   QaLabLatestReport,
@@ -44,7 +45,10 @@ import {
   type RuntimeParityCell,
   type RuntimeParityResult,
 } from "./runtime-parity.js";
-import { readQaBootstrapScenarioCatalog } from "./scenario-catalog.js";
+import {
+  readQaBootstrapScenarioCatalog,
+  type QaSeedScenarioWithSource,
+} from "./scenario-catalog.js";
 import { runScenarioFlow } from "./scenario-flow-runner.js";
 import {
   applyQaMergePatch,
@@ -536,6 +540,7 @@ export type QaSuiteSummaryJsonParams = {
   startedAt: Date;
   finishedAt: Date;
   metrics?: QaSuiteSummaryJson["metrics"];
+  evidence?: QaSuiteSummaryJson["evidence"];
   providerMode: QaProviderMode;
   primaryModel: string;
   alternateModel: string;
@@ -587,6 +592,7 @@ export function buildQaSuiteSummaryJson(params: QaSuiteSummaryJsonParams): QaSui
       failed: countQaSuiteFailedScenarios(params.scenarios),
     },
     ...(params.metrics ? { metrics: params.metrics } : {}),
+    ...(params.evidence ? { evidence: params.evidence } : {}),
     run: {
       startedAt: params.startedAt.toISOString(),
       finishedAt: params.finishedAt.toISOString(),
@@ -619,7 +625,7 @@ async function runQaRuntimeParitySuite(params: {
   claudeCliAuthMode?: QaCliBackendAuthMode;
   enabledPluginIds?: string[];
   concurrency: number;
-  selectedCatalogScenarios: ReturnType<typeof readQaBootstrapScenarioCatalog>["scenarios"];
+  selectedScenarios: ReturnType<typeof readQaBootstrapScenarioCatalog>["scenarios"];
   startLab?: QaSuiteStartLabFn;
   lab?: QaLabServerHandle;
   progressEnabled: boolean;
@@ -640,13 +646,11 @@ async function runQaRuntimeParitySuite(params: {
     id: params.transportId,
     state: lab.state,
   });
-  const liveScenarioOutcomes: QaLabScenarioOutcome[] = params.selectedCatalogScenarios.map(
-    (scenario) => ({
-      id: scenario.id,
-      name: scenario.title,
-      status: "pending",
-    }),
-  );
+  const liveScenarioOutcomes: QaLabScenarioOutcome[] = params.selectedScenarios.map((scenario) => ({
+    id: scenario.id,
+    name: scenario.title,
+    status: "pending",
+  }));
   lab.setScenarioRun({
     kind: "suite",
     status: "running",
@@ -656,13 +660,13 @@ async function runQaRuntimeParitySuite(params: {
 
   try {
     const scenarios = await mapQaSuiteWithConcurrency(
-      params.selectedCatalogScenarios,
+      params.selectedScenarios,
       params.concurrency,
       async (scenario, index): Promise<QaSuiteScenarioResult> => {
         const scenarioIdForLog = sanitizeQaSuiteProgressValue(scenario.id);
         writeQaSuiteProgress(
           params.progressEnabled,
-          `runtime pair start (${index + 1}/${params.selectedCatalogScenarios.length}): ${scenarioIdForLog}`,
+          `runtime pair start (${index + 1}/${params.selectedScenarios.length}): ${scenarioIdForLog}`,
         );
         liveScenarioOutcomes[index] = {
           id: scenario.id,
@@ -770,7 +774,7 @@ async function runQaRuntimeParitySuite(params: {
         });
         writeQaSuiteProgress(
           params.progressEnabled,
-          `runtime pair ${result.status} (${index + 1}/${params.selectedCatalogScenarios.length}): ${scenarioIdForLog}`,
+          `runtime pair ${result.status} (${index + 1}/${params.selectedScenarios.length}): ${scenarioIdForLog}`,
         );
         return result;
       },
@@ -785,6 +789,7 @@ async function runQaRuntimeParitySuite(params: {
       startedAt: params.startedAt,
       finishedAt,
       scenarios,
+      scenarioDefinitions: params.selectedScenarios,
       transport,
       providerMode: params.providerMode,
       primaryModel: params.primaryModel,
@@ -793,7 +798,7 @@ async function runQaRuntimeParitySuite(params: {
       concurrency: params.concurrency,
       scenarioIds:
         params.scenarioIds && params.scenarioIds.length > 0
-          ? params.selectedCatalogScenarios.map((scenario) => scenario.id)
+          ? params.selectedScenarios.map((scenario) => scenario.id)
           : undefined,
       runtimePair: params.runtimePair,
     });
@@ -829,6 +834,7 @@ async function writeQaSuiteArtifacts(params: {
   startedAt: Date;
   finishedAt: Date;
   scenarios: QaSuiteScenarioResult[];
+  scenarioDefinitions?: readonly QaSeedScenarioWithSource[];
   metrics?: QaSuiteSummaryJson["metrics"];
   transport: QaTransportAdapter;
   // Reuse the canonical QaProviderMode union instead of re-declaring it
@@ -844,6 +850,8 @@ async function writeQaSuiteArtifacts(params: {
   scenarioIds?: readonly string[];
   runtimePair?: [RuntimeId, RuntimeId];
 }) {
+  const reportPath = path.join(params.outputDir, "qa-suite-report.md");
+  const summaryPath = path.join(params.outputDir, "qa-suite-summary.json");
   const report = renderQaMarkdownReport({
     title: "OpenClaw QA Scenario Suite",
     startedAt: params.startedAt,
@@ -857,12 +865,26 @@ async function writeQaSuiteArtifacts(params: {
     })) satisfies QaReportScenario[],
     notes: createQaSuiteReportNotes(params),
   });
-  const reportPath = path.join(params.outputDir, "qa-suite-report.md");
-  const summaryPath = path.join(params.outputDir, "qa-suite-summary.json");
+  const evidence =
+    params.scenarioDefinitions && params.scenarioDefinitions.length > 0
+      ? buildQaSuiteEvidenceSummary({
+          artifactPaths: [
+            { kind: "summary", path: path.basename(summaryPath) },
+            { kind: "report", path: path.basename(reportPath) },
+          ],
+          channelId: params.transport.id,
+          env: process.env,
+          generatedAt: params.finishedAt.toISOString(),
+          primaryModel: params.primaryModel,
+          providerMode: params.providerMode,
+          scenarioDefinitions: params.scenarioDefinitions,
+          scenarioResults: params.scenarios,
+        })
+      : undefined;
   await fs.writeFile(reportPath, report, "utf8");
   await fs.writeFile(
     summaryPath,
-    `${JSON.stringify(buildQaSuiteSummaryJson(params), null, 2)}\n`,
+    `${JSON.stringify(buildQaSuiteSummaryJson({ ...params, evidence }), null, 2)}\n`,
     "utf8",
   );
   return { report, reportPath, summaryPath };
@@ -1018,7 +1040,7 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
       : isQaFastModeEnabled({ primaryModel, alternateModel });
   const outputDir = await resolveQaSuiteOutputDir(repoRoot, params?.outputDir);
   const catalog = readQaBootstrapScenarioCatalog();
-  const selectedCatalogScenarios = selectQaSuiteScenarios({
+  const selectedScenarios = selectQaSuiteScenarios({
     scenarios: catalog.scenarios,
     scenarioIds: params?.scenarioIds,
     providerMode,
@@ -1027,28 +1049,28 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
   });
   const enabledPluginIds = [
     ...new Set([
-      ...collectQaSuitePluginIds(selectedCatalogScenarios),
+      ...collectQaSuitePluginIds(selectedScenarios),
       ...(params?.enabledPluginIds ?? []).map((pluginId) => pluginId.trim()).filter(Boolean),
       ...(params?.forcedRuntime && params.forcedRuntime !== "openclaw"
         ? [params.forcedRuntime]
         : []),
     ]),
   ];
-  const gatewayConfigPatch = collectQaSuiteGatewayConfigPatch(selectedCatalogScenarios);
-  const gatewayRuntimeOptions = collectQaSuiteGatewayRuntimeOptions(selectedCatalogScenarios);
+  const gatewayConfigPatch = collectQaSuiteGatewayConfigPatch(selectedScenarios);
+  const gatewayRuntimeOptions = collectQaSuiteGatewayRuntimeOptions(selectedScenarios);
   const concurrency = normalizeQaSuiteConcurrency(
     params?.concurrency,
-    selectedCatalogScenarios.length,
+    selectedScenarios.length,
     defaultQaSuiteConcurrencyForTransport(transportId),
   );
   const progressEnabled = shouldLogQaSuiteProgress();
   const gatewayHeapCheckpointsEnabled = shouldCaptureGatewayHeapCheckpoints();
   writeQaSuiteProgress(
     progressEnabled,
-    `run start: scenarios=${selectedCatalogScenarios.length} concurrency=${concurrency} transport=${transportId}`,
+    `run start: scenarios=${selectedScenarios.length} concurrency=${concurrency} transport=${transportId}`,
   );
   const useIsolatedScenarioWorkers = shouldRunQaSuiteWithIsolatedScenarioWorkers({
-    scenarios: selectedCatalogScenarios,
+    scenarios: selectedScenarios,
     concurrency,
     lab: params?.lab,
     startLab: params?.startLab,
@@ -1068,7 +1090,7 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
       claudeCliAuthMode: params.claudeCliAuthMode,
       enabledPluginIds: params.enabledPluginIds,
       concurrency,
-      selectedCatalogScenarios,
+      selectedScenarios,
       startLab: params.startLab,
       lab: params.lab,
       progressEnabled,
@@ -1092,13 +1114,11 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
       id: transportId,
       state: lab.state,
     });
-    const liveScenarioOutcomes: QaLabScenarioOutcome[] = selectedCatalogScenarios.map(
-      (scenario) => ({
-        id: scenario.id,
-        name: scenario.title,
-        status: "pending",
-      }),
-    );
+    const liveScenarioOutcomes: QaLabScenarioOutcome[] = selectedScenarios.map((scenario) => ({
+      id: scenario.id,
+      name: scenario.title,
+      status: "pending",
+    }));
     const updateScenarioRun = () =>
       lab.setScenarioRun({
         kind: "suite",
@@ -1107,13 +1127,16 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
         scenarios: [...liveScenarioOutcomes],
       });
     const completedScenarioResults: Array<QaSuiteScenarioResult | undefined> = Array.from({
-      length: selectedCatalogScenarios.length,
+      length: selectedScenarios.length,
     });
     let artifactWriteQueue = Promise.resolve();
     const writePartialArtifacts = () => {
       const partialScenarios = completedScenarioResults.filter(
         (scenario): scenario is QaSuiteScenarioResult => scenario !== undefined,
       );
+      const completedScenarioDefinitions = completedScenarioResults.flatMap((scenario, index) =>
+        scenario === undefined ? [] : [selectedScenarios[index]],
+      );
       if (partialScenarios.length === 0) {
         return;
       }
@@ -1125,6 +1148,7 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
             startedAt,
             finishedAt: partialFinishedAt,
             scenarios: partialScenarios,
+            scenarioDefinitions: completedScenarioDefinitions,
             transport,
             providerMode,
             primaryModel,
@@ -1134,7 +1158,7 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
             isolatedWorkers: true,
             scenarioIds:
               params?.scenarioIds && params.scenarioIds.length > 0
-                ? selectedCatalogScenarios.map((scenario) => scenario.id)
+                ? selectedScenarios.map((scenario) => scenario.id)
                 : undefined,
           });
           lab.setLatestReport({
@@ -1156,13 +1180,13 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
       const workerStartStaggerMs = resolveQaSuiteWorkerStartStaggerMs(concurrency);
       writeQaSuiteProgress(progressEnabled, `scenario start stagger=${workerStartStaggerMs}ms`);
       const scenarios: QaSuiteScenarioResult[] = await mapQaSuiteWithConcurrency(
-        selectedCatalogScenarios,
+        selectedScenarios,
         concurrency,
         async (scenario, index): Promise<QaSuiteScenarioResult> => {
           const scenarioIdForLog = sanitizeQaSuiteProgressValue(scenario.id);
           writeQaSuiteProgress(
             progressEnabled,
-            `scenario start (${index + 1}/${selectedCatalogScenarios.length}): ${scenarioIdForLog}`,
+            `scenario start (${index + 1}/${selectedScenarios.length}): ${scenarioIdForLog}`,
           );
           liveScenarioOutcomes[index] = {
             id: scenario.id,
@@ -1213,7 +1237,7 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
             updateScenarioRun();
             writeQaSuiteProgress(
               progressEnabled,
-              `scenario ${scenarioResult.status} (${index + 1}/${selectedCatalogScenarios.length}): ${scenarioIdForLog}`,
+              `scenario ${scenarioResult.status} (${index + 1}/${selectedScenarios.length}): ${scenarioIdForLog}`,
             );
             completedScenarioResults[index] = scenarioResult;
             writePartialArtifacts();
@@ -1244,7 +1268,7 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
             updateScenarioRun();
             writeQaSuiteProgress(
               progressEnabled,
-              `scenario fail (${index + 1}/${selectedCatalogScenarios.length}): ${scenarioIdForLog}`,
+              `scenario fail (${index + 1}/${selectedScenarios.length}): ${scenarioIdForLog}`,
             );
             completedScenarioResults[index] = scenarioResult;
             writePartialArtifacts();
@@ -1268,6 +1292,7 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
         startedAt,
         finishedAt,
         scenarios,
+        scenarioDefinitions: selectedScenarios,
         transport,
         providerMode,
         primaryModel,
@@ -1283,7 +1308,7 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
         // distinguishable from an explicit all-scenarios selection.
         scenarioIds:
           params?.scenarioIds && params.scenarioIds.length > 0
-            ? selectedCatalogScenarios.map((scenario) => scenario.id)
+            ? selectedScenarios.map((scenario) => scenario.id)
             : undefined,
       });
       lab.setLatestReport({
@@ -1399,13 +1424,11 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
     });
     await sleep(1_000);
     const scenarios: QaSuiteScenarioResult[] = [];
-    const liveScenarioOutcomes: QaLabScenarioOutcome[] = selectedCatalogScenarios.map(
-      (scenario) => ({
-        id: scenario.id,
-        name: scenario.title,
-        status: "pending",
-      }),
-    );
+    const liveScenarioOutcomes: QaLabScenarioOutcome[] = selectedScenarios.map((scenario) => ({
+      id: scenario.id,
+      name: scenario.title,
+      status: "pending",
+    }));
 
     lab.setScenarioRun({
       kind: "suite",
@@ -1443,11 +1466,11 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
       }
     };
     await captureGatewayHeapCheckpoint("suite-start");
-    for (const [index, scenario] of selectedCatalogScenarios.entries()) {
+    for (const [index, scenario] of selectedScenarios.entries()) {
       const scenarioIdForLog = sanitizeQaSuiteProgressValue(scenario.id);
       writeQaSuiteProgress(
         progressEnabled,
-        `scenario start (${index + 1}/${selectedCatalogScenarios.length}): ${scenarioIdForLog}`,
+        `scenario start (${index + 1}/${selectedScenarios.length}): ${scenarioIdForLog}`,
       );
       sampleGatewayProcessRss(`scenario:${scenario.id}:start`);
       liveScenarioOutcomes[index] = {
@@ -1468,7 +1491,7 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
       scenarios.push(result);
       writeQaSuiteProgress(
         progressEnabled,
-        `scenario ${result.status} (${index + 1}/${selectedCatalogScenarios.length}): ${scenarioIdForLog}`,
+        `scenario ${result.status} (${index + 1}/${selectedScenarios.length}): ${scenarioIdForLog}`,
       );
       liveScenarioOutcomes[index] = {
         id: scenario.id,
@@ -1490,7 +1513,7 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
     const runtimeParityCell =
       params?.captureRuntimeParityCell &&
       params.forcedRuntime &&
-      selectedCatalogScenarios.length === 1 &&
+      selectedScenarios.length === 1 &&
       scenarios.length > 0
         ? await captureRuntimeParityCell({
             runtime: params.forcedRuntime,
@@ -1529,6 +1552,7 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
       finishedAt,
       scenarios,
       metrics,
+      scenarioDefinitions: selectedScenarios,
       transport,
       providerMode,
       primaryModel,
@@ -1540,7 +1564,7 @@ export async function runQaSuite(params?: QaSuiteRunParams): Promise<QaSuiteResu
       // the concurrent-path writeQaSuiteArtifacts call above.
       scenarioIds:
         params?.scenarioIds && params.scenarioIds.length > 0
-          ? selectedCatalogScenarios.map((scenario) => scenario.id)
+          ? selectedScenarios.map((scenario) => scenario.id)
           : undefined,
     });
     const latestReport = {

extensions/telegram/src/bot-core.ts

@@ -28,6 +28,12 @@ import type { TelegramBotDeps } from "./bot-deps.js";
 import { registerTelegramHandlers } from "./bot-handlers.runtime.js";
 import { createTelegramMessageProcessor } from "./bot-message.js";
 import { registerTelegramNativeCommands } from "./bot-native-commands.js";
+import {
+  getTelegramSpooledReplayDeferredParticipant,
+  isTelegramSpooledReplayUpdate,
+  runWithTelegramUpdateProcessingFrame,
+  TelegramSpooledReplayProcessingError,
+} from "./bot-processing-outcome.js";
 import { createTelegramUpdateTracker } from "./bot-update-tracker.js";
 import type { TelegramUpdateKeyContext } from "./bot-updates.js";
 import { resolveDefaultAgentId } from "./bot.agent.runtime.js";
@@ -212,7 +218,29 @@ export function createTelegramBotCore(
       return;
     }
     try {
-      await next();
+      const { result } = await runWithTelegramUpdateProcessingFrame(async () => {
+        await next();
+      });
+      const deferredWork = getTelegramSpooledReplayDeferredParticipant();
+      if (deferredWork) {
+        void deferredWork.task
+          .then((deferredResult) => {
+            updateTracker.finishUpdate(begin.update, {
+              completed: deferredResult.kind !== "failed-retryable",
+            });
+          })
+          .catch(() => {
+            updateTracker.finishUpdate(begin.update, { completed: false });
+          });
+        return;
+      }
+      if (result?.kind === "failed-retryable") {
+        if (isTelegramSpooledReplayUpdate(ctx.update)) {
+          throw new TelegramSpooledReplayProcessingError(result.error);
+        }
+        updateTracker.finishUpdate(begin.update, { completed: true });
+        return;
+      }
       updateTracker.finishUpdate(begin.update, { completed: true });
     } catch (error) {
       updateTracker.finishUpdate(begin.update, { completed: false });

extensions/telegram/src/bot-handlers.runtime.ts

@@ -72,6 +72,12 @@ import type {
 } from "./bot-message-context.types.js";
 import { parseTelegramNativeCommandCallbackData } from "./bot-native-commands.js";
 import type { RegisterTelegramHandlerParams } from "./bot-native-commands.js";
+import {
+  createTelegramSpooledReplayDeferredParticipant,
+  isTelegramSpooledReplayUpdate,
+  type TelegramMessageProcessingResult,
+  type TelegramSpooledReplayDeferredParticipant,
+} from "./bot-processing-outcome.js";
 import {
   MEDIA_GROUP_TIMEOUT_MS,
   type MediaGroupEntry,
@@ -133,6 +139,7 @@ import {
   claimTelegramMessageDispatchReplay,
   commitTelegramMessageDispatchReplay,
   createTelegramMessageDispatchReplayGuard,
+  forgetTelegramMessageDispatchReplay,
   releaseTelegramMessageDispatchReplay,
 } from "./message-dispatch-dedupe.js";
 import {
@@ -145,6 +152,10 @@ import {
   type ProviderInfo,
 } from "./model-buttons.js";
 import { parseTelegramOpaqueCallbackData } from "./native-command-callback-data.js";
+import {
+  isTelegramEditTargetMissingError,
+  isTelegramMessageHasNoTextError,
+} from "./network-errors.js";
 import { buildInlineKeyboard } from "./send.js";
 
 export const registerTelegramHandlers = ({
@@ -204,6 +215,7 @@ export const registerTelegramHandlers = ({
     groupConfig?: TelegramGroupConfig;
     topicConfig?: TelegramTopicConfig;
     dispatchDedupeKeys: string[];
+    spooledReplayParticipants: TelegramSpooledReplayDeferredParticipant[];
   };
 
   const mediaGroupBuffer = new Map<string, BufferedMediaGroupEntry>();
@@ -223,6 +235,7 @@ export const registerTelegramHandlers = ({
     messages: Array<{ msg: Message; ctx: TelegramContext; receivedAtMs: number }>;
     promptContextMinTimestampMs?: number;
     dispatchDedupeKeys: string[];
+    spooledReplayParticipants: TelegramSpooledReplayDeferredParticipant[];
     timer: ReturnType<typeof setTimeout>;
   };
   const textFragmentBuffer = new Map<string, TextFragmentEntry>();
@@ -257,6 +270,26 @@ export const registerTelegramHandlers = ({
     threadId?: number;
     promptContextMinTimestampMs?: number;
     dispatchDedupeKeys: string[];
+    spooledReplayParticipant?: TelegramSpooledReplayDeferredParticipant;
+  };
+  const resolveTelegramDebounceEntryMs = (entry: TelegramDebounceEntry): number =>
+    entry.debounceLane === "forward" ? FORWARD_BURST_DEBOUNCE_MS : debounceMs;
+  const shouldDebounceTelegramEntry = (entry: TelegramDebounceEntry): boolean => {
+    const text = getTelegramTextParts(entry.msg).text;
+    const hasDebounceableText = shouldDebounceTextInbound({
+      text,
+      cfg,
+      commandOptions: { botUsername: entry.botUsername },
+    });
+    if (entry.debounceLane === "forward") {
+      // Forwarded bursts often split text + media into adjacent updates.
+      // Debounce media-only forward entries too so they can coalesce.
+      return hasDebounceableText || entry.allMedia.length > 0;
+    }
+    if (!hasDebounceableText) {
+      return false;
+    }
+    return entry.allMedia.length === 0;
   };
   const normalizePromptContextMinTimestampMs = (timestampMs?: number) =>
     typeof timestampMs === "number" && Number.isFinite(timestampMs) ? timestampMs : undefined;
@@ -295,6 +328,32 @@ export const registerTelegramHandlers = ({
       keys,
     });
   };
+  const forgetDispatchDedupeKeys = async (keys: readonly string[]) => {
+    await forgetTelegramMessageDispatchReplay({
+      guard: messageDispatchReplayGuard,
+      keys,
+    });
+  };
+  const buildFailedProcessingResult = (error: unknown): TelegramMessageProcessingResult => ({
+    kind: "failed-retryable",
+    error,
+  });
+  const settleSpooledReplayParticipants = (
+    participants: readonly TelegramSpooledReplayDeferredParticipant[],
+    result: TelegramMessageProcessingResult,
+  ) => {
+    for (const participant of new Set(participants)) {
+      participant.settle(result);
+    }
+  };
+  const createSpooledReplayParticipantForBufferedWork = (
+    key: string,
+  ): TelegramSpooledReplayDeferredParticipant | undefined =>
+    createTelegramSpooledReplayDeferredParticipant(key) ?? undefined;
+  const spooledReplayOptions = (
+    participants: readonly TelegramSpooledReplayDeferredParticipant[],
+  ): Pick<TelegramMessageContextOptions, "spooledReplay"> =>
+    participants.length > 0 ? { spooledReplay: true } : {};
   const claimMessageDispatchDedupe = async (
     msg: Message,
   ): Promise<{ process: true; keys: string[] } | { process: false }> => {
@@ -475,84 +534,96 @@ export const registerTelegramHandlers = ({
   const inboundDebouncer = createInboundDebouncer<TelegramDebounceEntry>({
     debounceMs,
     serializeImmediate: true,
-    resolveDebounceMs: (entry) =>
-      entry.debounceLane === "forward" ? FORWARD_BURST_DEBOUNCE_MS : debounceMs,
+    resolveDebounceMs: resolveTelegramDebounceEntryMs,
     buildKey: (entry) => entry.debounceKey,
-    shouldDebounce: (entry) => {
-      const text = getTelegramTextParts(entry.msg).text;
-      const hasDebounceableText = shouldDebounceTextInbound({
-        text,
-        cfg,
-        commandOptions: { botUsername: entry.botUsername },
-      });
-      if (entry.debounceLane === "forward") {
-        // Forwarded bursts often split text + media into adjacent updates.
-        // Debounce media-only forward entries too so they can coalesce.
-        return hasDebounceableText || entry.allMedia.length > 0;
-      }
-      if (!hasDebounceableText) {
-        return false;
-      }
-      return entry.allMedia.length === 0;
-    },
+    shouldDebounce: shouldDebounceTelegramEntry,
     onFlush: async (entries) => {
+      const spooledReplayParticipants = entries
+        .map((entry) => entry.spooledReplayParticipant)
+        .filter(
+          (participant): participant is TelegramSpooledReplayDeferredParticipant =>
+            participant !== undefined,
+        );
       const last = entries.at(-1);
       if (!last) {
         return;
       }
-      if (entries.length === 1) {
-        await processMessageWithReplyChain({
-          ctx: last.ctx,
-          msg: last.msg,
-          allMedia: last.allMedia,
-          storeAllowFrom: last.storeAllowFrom,
-          options: {
-            receivedAtMs: last.receivedAtMs,
-            ingressBuffer: "inbound-debounce",
-            ...promptContextBoundaryOptions(last.promptContextMinTimestampMs),
-          },
-          dispatchDedupeKeys: last.dispatchDedupeKeys,
+      try {
+        if (entries.length === 1) {
+          const result = await processMessageWithReplyChain({
+            ctx: last.ctx,
+            msg: last.msg,
+            allMedia: last.allMedia,
+            storeAllowFrom: last.storeAllowFrom,
+            options: {
+              receivedAtMs: last.receivedAtMs,
+              ingressBuffer: "inbound-debounce",
+              ...promptContextBoundaryOptions(last.promptContextMinTimestampMs),
+              ...spooledReplayOptions(spooledReplayParticipants),
+            },
+            dispatchDedupeKeys: last.dispatchDedupeKeys,
+          });
+          settleSpooledReplayParticipants(spooledReplayParticipants, result);
+          return;
+        }
+        const combinedText = entries
+          .map((entry) => getTelegramTextParts(entry.msg).text)
+          .filter(Boolean)
+          .join("\n");
+        const combinedMedia = entries.flatMap((entry) => entry.allMedia);
+        if (!combinedText.trim() && combinedMedia.length === 0) {
+          settleSpooledReplayParticipants(spooledReplayParticipants, { kind: "skipped" });
+          return;
+        }
+        const first = entries[0];
+        const promptContextMinTimestampMs = latestPromptContextMinTimestampMs(
+          ...entries.map((entry) => entry.promptContextMinTimestampMs),
+        );
+        const baseCtx = first.ctx;
+        const syntheticMessage = buildSyntheticTextMessage({
+          base: first.msg,
+          text: combinedText,
+          date: last.msg.date ?? first.msg.date,
         });
-        return;
+        const messageIdOverride = last.msg.message_id ? String(last.msg.message_id) : undefined;
+        const syntheticCtx = buildSyntheticContext(baseCtx, syntheticMessage);
+        const result = await processMessageWithReplyChain({
+          ctx: syntheticCtx,
+          msg: syntheticMessage,
+          allMedia: combinedMedia,
+          storeAllowFrom: first.storeAllowFrom,
+          options: {
+            ...(messageIdOverride ? { messageIdOverride } : {}),
+            receivedAtMs: first.receivedAtMs,
+            ingressBuffer: "inbound-debounce",
+            ...promptContextBoundaryOptions(promptContextMinTimestampMs),
+            ...spooledReplayOptions(spooledReplayParticipants),
+          },
+          dispatchDedupeKeys: mergeDispatchDedupeKeys(
+            ...entries.map((entry) => entry.dispatchDedupeKeys),
+          ),
+        });
+        settleSpooledReplayParticipants(spooledReplayParticipants, result);
+      } catch (err) {
+        settleSpooledReplayParticipants(
+          spooledReplayParticipants,
+          buildFailedProcessingResult(err),
+        );
+        throw err;
       }
-      const combinedText = entries
-        .map((entry) => getTelegramTextParts(entry.msg).text)
-        .filter(Boolean)
-        .join("\n");
-      const combinedMedia = entries.flatMap((entry) => entry.allMedia);
-      if (!combinedText.trim() && combinedMedia.length === 0) {
-        return;
-      }
-      const first = entries[0];
-      const promptContextMinTimestampMs = latestPromptContextMinTimestampMs(
-        ...entries.map((entry) => entry.promptContextMinTimestampMs),
-      );
-      const baseCtx = first.ctx;
-      const syntheticMessage = buildSyntheticTextMessage({
-        base: first.msg,
-        text: combinedText,
-        date: last.msg.date ?? first.msg.date,
-      });
-      const messageIdOverride = last.msg.message_id ? String(last.msg.message_id) : undefined;
-      const syntheticCtx = buildSyntheticContext(baseCtx, syntheticMessage);
-      await processMessageWithReplyChain({
-        ctx: syntheticCtx,
-        msg: syntheticMessage,
-        allMedia: combinedMedia,
-        storeAllowFrom: first.storeAllowFrom,
-        options: {
-          ...(messageIdOverride ? { messageIdOverride } : {}),
-          receivedAtMs: first.receivedAtMs,
-          ingressBuffer: "inbound-debounce",
-          ...promptContextBoundaryOptions(promptContextMinTimestampMs),
-        },
-        dispatchDedupeKeys: mergeDispatchDedupeKeys(
-          ...entries.map((entry) => entry.dispatchDedupeKeys),
-        ),
-      });
     },
     onError: (err, items) => {
+      const spooledReplayParticipants = items
+        .map((item) => item.spooledReplayParticipant)
+        .filter(
+          (participant): participant is TelegramSpooledReplayDeferredParticipant =>
+            participant !== undefined,
+        );
+      settleSpooledReplayParticipants(spooledReplayParticipants, buildFailedProcessingResult(err));
       runtime.error?.(danger(`telegram debounce flush failed: ${String(err)}`));
+      if (spooledReplayParticipants.length > 0) {
+        return;
+      }
       const chatId = items[0]?.msg.chat.id;
       if (chatId != null) {
         const threadId = items[0]?.msg.message_thread_id;
@@ -568,6 +639,15 @@ export const registerTelegramHandlers = ({
       }
     },
     onCancel: (items) => {
+      settleSpooledReplayParticipants(
+        items
+          .map((item) => item.spooledReplayParticipant)
+          .filter(
+            (participant): participant is TelegramSpooledReplayDeferredParticipant =>
+              participant !== undefined,
+          ),
+        { kind: "skipped" },
+      );
       releaseDispatchDedupeKeys(
         mergeDispatchDedupeKeys(...items.map((item) => item.dispatchDedupeKeys)),
       );
@@ -808,6 +888,7 @@ export const registerTelegramHandlers = ({
       const primaryEntry = captionMsg ?? entry.messages[0];
       if (!primaryEntry) {
         releaseDispatchDedupeKeys(entry.dispatchDedupeKeys);
+        settleSpooledReplayParticipants(entry.spooledReplayParticipants, { kind: "skipped" });
         return;
       }
 
@@ -828,6 +909,7 @@ export const registerTelegramHandlers = ({
         })
       ) {
         releaseDispatchDedupeKeys(entry.dispatchDedupeKeys);
+        settleSpooledReplayParticipants(entry.spooledReplayParticipants, { kind: "skipped" });
         return;
       }
 
@@ -882,16 +964,24 @@ export const registerTelegramHandlers = ({
         }).catch(() => {});
       }
 
-      await processMessageWithReplyChain({
+      const result = await processMessageWithReplyChain({
         ctx: primaryEntry.ctx,
         msg: primaryEntry.msg,
         allMedia,
         storeAllowFrom: entry.storeAllowFrom,
-        options: promptContextBoundaryOptions(entry.promptContextMinTimestampMs),
+        options: {
+          ...promptContextBoundaryOptions(entry.promptContextMinTimestampMs),
+          ...spooledReplayOptions(entry.spooledReplayParticipants),
+        },
         dispatchDedupeKeys: entry.dispatchDedupeKeys,
       });
+      settleSpooledReplayParticipants(entry.spooledReplayParticipants, result);
     } catch (err) {
       releaseDispatchDedupeKeys(entry.dispatchDedupeKeys, err);
+      settleSpooledReplayParticipants(
+        entry.spooledReplayParticipants,
+        buildFailedProcessingResult(err),
+      );
       runtime.error?.(danger(`media group handler failed: ${String(err)}`));
     }
   };
@@ -904,12 +994,14 @@ export const registerTelegramHandlers = ({
       const last = entry.messages.at(-1);
       if (!first || !last) {
         releaseDispatchDedupeKeys(entry.dispatchDedupeKeys);
+        settleSpooledReplayParticipants(entry.spooledReplayParticipants, { kind: "skipped" });
         return;
       }
 
       const combinedText = entry.messages.map((m) => m.msg.text ?? "").join("");
       if (!combinedText.trim()) {
         releaseDispatchDedupeKeys(entry.dispatchDedupeKeys);
+        settleSpooledReplayParticipants(entry.spooledReplayParticipants, { kind: "skipped" });
         return;
       }
 
@@ -923,7 +1015,7 @@ export const registerTelegramHandlers = ({
       const baseCtx = first.ctx;
 
       const syntheticCtx = buildSyntheticContext(baseCtx, syntheticMessage);
-      await processMessageWithReplyChain({
+      const result = await processMessageWithReplyChain({
         ctx: syntheticCtx,
         msg: syntheticMessage,
         allMedia: [],
@@ -933,11 +1025,17 @@ export const registerTelegramHandlers = ({
           receivedAtMs: first.receivedAtMs,
           ingressBuffer: "text-fragment",
           ...promptContextBoundaryOptions(entry.promptContextMinTimestampMs),
+          ...spooledReplayOptions(entry.spooledReplayParticipants),
         },
         dispatchDedupeKeys: entry.dispatchDedupeKeys,
       });
+      settleSpooledReplayParticipants(entry.spooledReplayParticipants, result);
     } catch (err) {
       releaseDispatchDedupeKeys(entry.dispatchDedupeKeys, err);
+      settleSpooledReplayParticipants(
+        entry.spooledReplayParticipants,
+        buildFailedProcessingResult(err),
+      );
       runtime.error?.(danger(`text fragment handler failed: ${String(err)}`));
     }
   };
@@ -1121,8 +1219,15 @@ export const registerTelegramHandlers = ({
     storeAllowFrom: string[];
     options?: TelegramMessageContextOptions;
     dispatchDedupeKeys?: string[];
-  }) => {
+  }): Promise<TelegramMessageProcessingResult> => {
     let dispatchDedupeCommitted = false;
+    let dispatchDedupeRollbackAttempted = false;
+    const spooledReplay =
+      params.options?.spooledReplay === true || isTelegramSpooledReplayUpdate(params.ctx.update);
+    const forgetCommittedDispatchDedupeKeys = async () => {
+      dispatchDedupeRollbackAttempted = true;
+      await forgetDispatchDedupeKeys(params.dispatchDedupeKeys ?? []);
+    };
     try {
       const replyChainNodes = await buildReplyChainForMessage(params.msg);
       const { replyMedia, replyChain } = await resolveReplyMediaForChain(
@@ -1134,7 +1239,7 @@ export const registerTelegramHandlers = ({
         replyChainNodes,
         params.options,
       );
-      const dispatched = await processMessage(
+      const result = await processMessage(
         params.ctx,
         params.allMedia,
         params.storeAllowFrom,
@@ -1149,11 +1254,18 @@ export const registerTelegramHandlers = ({
           },
         },
       );
-      if (!dispatched && !dispatchDedupeCommitted) {
+      if (result.kind === "completed" && !dispatchDedupeCommitted) {
+        await commitDispatchDedupeKeys(params.dispatchDedupeKeys ?? []);
+      } else if (result.kind === "failed-retryable" && dispatchDedupeCommitted && spooledReplay) {
+        await forgetCommittedDispatchDedupeKeys();
+      } else if (result.kind !== "completed" && !dispatchDedupeCommitted) {
         releaseDispatchDedupeKeys(params.dispatchDedupeKeys ?? []);
       }
+      return result;
     } catch (err) {
-      if (!dispatchDedupeCommitted) {
+      if (dispatchDedupeCommitted && spooledReplay && !dispatchDedupeRollbackAttempted) {
+        await forgetCommittedDispatchDedupeKeys();
+      } else if (!dispatchDedupeCommitted) {
         releaseDispatchDedupeKeys(params.dispatchDedupeKeys ?? [], err);
       }
       throw err;
@@ -1303,11 +1415,8 @@ export const registerTelegramHandlers = ({
     }
   }
 
-  const TELEGRAM_PERMANENT_CALLBACK_EDIT_ERROR_RE =
-    /400:\s*Bad Request:\s*message to edit not found|400:\s*Bad Request:\s*there is no text in the message to edit|MESSAGE_ID_INVALID|400:\s*Bad Request:\s*message can't be edited/i;
-
   const isPermanentTelegramCallbackEditError = (err: unknown): boolean =>
-    TELEGRAM_PERMANENT_CALLBACK_EDIT_ERROR_RE.test(String(err));
+    isTelegramEditTargetMissingError(err) || isTelegramMessageHasNoTextError(err);
 
   const resolveTelegramEventAuthorizationContext = async (params: {
     chatId: number;
@@ -1726,6 +1835,12 @@ export const registerTelegramHandlers = ({
             existing.messages.length + 1 <= TELEGRAM_TEXT_FRAGMENT_MAX_PARTS &&
             nextTotalChars <= TELEGRAM_TEXT_FRAGMENT_MAX_TOTAL_CHARS
           ) {
+            const spooledReplayParticipant = createSpooledReplayParticipantForBufferedWork(
+              `text-fragment:${key}:${msg.message_id}`,
+            );
+            if (spooledReplayParticipant) {
+              existing.spooledReplayParticipants.push(spooledReplayParticipant);
+            }
             existing.messages.push({ msg, ctx, receivedAtMs: nowMs });
             existing.promptContextMinTimestampMs = latestPromptContextMinTimestampMs(
               existing.promptContextMinTimestampMs,
@@ -1748,10 +1863,14 @@ export const registerTelegramHandlers = ({
 
       const shouldStart = text.length >= TELEGRAM_TEXT_FRAGMENT_START_THRESHOLD_CHARS;
       if (shouldStart) {
+        const spooledReplayParticipant = createSpooledReplayParticipantForBufferedWork(
+          `text-fragment:${key}:${msg.message_id}`,
+        );
         const entry: TextFragmentEntry = {
           key,
           messages: [{ msg, ctx, receivedAtMs: nowMs }],
           dispatchDedupeKeys,
+          spooledReplayParticipants: spooledReplayParticipant ? [spooledReplayParticipant] : [],
           ...promptContextBoundaryOptions(promptContextMinTimestampMs),
           timer: setTimeout(() => {}, TELEGRAM_TEXT_FRAGMENT_MAX_GAP_MS),
         };
@@ -1768,6 +1887,7 @@ export const registerTelegramHandlers = ({
         clearTimeout(existing.timer);
         textFragmentBuffer.delete(key);
         releaseDispatchDedupeKeys(existing.dispatchDedupeKeys);
+        settleSpooledReplayParticipants(existing.spooledReplayParticipants, { kind: "skipped" });
       }
     }
 
@@ -1778,6 +1898,12 @@ export const registerTelegramHandlers = ({
       const mediaGroupKey = `media:${chatId}:${threadId ?? "main"}:${mediaGroupId}`;
       const existing = mediaGroupBuffer.get(mediaGroupKey);
       if (existing) {
+        const spooledReplayParticipant = createSpooledReplayParticipantForBufferedWork(
+          `media-group:${mediaGroupKey}:${msg.message_id}`,
+        );
+        if (spooledReplayParticipant) {
+          existing.spooledReplayParticipants.push(spooledReplayParticipant);
+        }
         clearTimeout(existing.timer);
         existing.messages.push({ msg, ctx });
         existing.promptContextMinTimestampMs = latestPromptContextMinTimestampMs(
@@ -1795,6 +1921,9 @@ export const registerTelegramHandlers = ({
           });
         }, mediaGroupTimeoutMs);
       } else {
+        const spooledReplayParticipant = createSpooledReplayParticipantForBufferedWork(
+          `media-group:${mediaGroupKey}:${msg.message_id}`,
+        );
         const entry: BufferedMediaGroupEntry = {
           messages: [{ msg, ctx }],
           storeAllowFrom,
@@ -1808,6 +1937,7 @@ export const registerTelegramHandlers = ({
           groupConfig,
           topicConfig,
           dispatchDedupeKeys,
+          spooledReplayParticipants: spooledReplayParticipant ? [spooledReplayParticipant] : [],
           ...promptContextBoundaryOptions(promptContextMinTimestampMs),
           timer: setTimeout(() => {
             mediaGroupBuffer.delete(mediaGroupKey);
@@ -1927,7 +2057,7 @@ export const registerTelegramHandlers = ({
         );
       }
     }
-    await inboundDebouncer.enqueue({
+    const debounceEntry: TelegramDebounceEntry = {
       ctx,
       msg,
       allMedia,
@@ -1938,7 +2068,17 @@ export const registerTelegramHandlers = ({
       botUsername,
       ...promptContextBoundaryOptions(promptContextMinTimestampMs),
       dispatchDedupeKeys,
-    });
+    };
+    if (
+      debounceEntry.debounceKey &&
+      resolveTelegramDebounceEntryMs(debounceEntry) > 0 &&
+      shouldDebounceTelegramEntry(debounceEntry)
+    ) {
+      debounceEntry.spooledReplayParticipant = createSpooledReplayParticipantForBufferedWork(
+        `inbound-debounce:${debounceEntry.debounceKey}`,
+      );
+    }
+    await inboundDebouncer.enqueue(debounceEntry);
   };
   bot.on("callback_query", async (ctx) => {
     const callback = ctx.callbackQuery;

extensions/telegram/src/bot-message-context.types.ts

@@ -25,6 +25,7 @@ export type TelegramMessageContextOptions = {
   receivedAtMs?: number;
   ingressBuffer?: "inbound-debounce" | "text-fragment";
   promptContextMinTimestampMs?: number;
+  spooledReplay?: boolean;
 };
 
 export type TelegramPromptContextEntry = NonNullable<

extensions/telegram/src/bot-message-dispatch.test.ts

@@ -523,10 +523,12 @@ describe("dispatchTelegramMessage draft streaming", () => {
     telegramDeps?: TelegramBotDeps;
     bot?: Bot;
     replyToMode?: Parameters<typeof dispatchTelegramMessage>[0]["replyToMode"];
+    retryDispatchErrors?: boolean;
+    suppressFailureFallback?: boolean;
     textLimit?: number;
   }) {
     const bot = params.bot ?? createBot();
-    await dispatchTelegramMessage({
+    return await dispatchTelegramMessage({
       context: params.context,
       bot,
       cfg: params.cfg ?? {},
@@ -537,6 +539,8 @@ describe("dispatchTelegramMessage draft streaming", () => {
       telegramCfg: params.telegramCfg ?? {},
       telegramDeps: params.telegramDeps ?? telegramDepsForTest,
       opts: { token: "token" },
+      retryDispatchErrors: params.retryDispatchErrors,
+      suppressFailureFallback: params.suppressFailureFallback,
     });
   }
 
@@ -2401,6 +2405,42 @@ describe("dispatchTelegramMessage draft streaming", () => {
     expect(deliverReplies).toHaveBeenCalledTimes(1);
   });
 
+  it("returns retryable when spooled replay suppresses fallback after non-silent delivery skip", async () => {
+    dispatchReplyWithBufferedBlockDispatcher.mockImplementation(async ({ dispatcherOptions }) => {
+      dispatcherOptions.onSkip?.({ text: "final answer" }, { kind: "final", reason: "empty" });
+      return { queuedFinal: false };
+    });
+
+    const result = await dispatchWithContext({
+      context: createContext(),
+      retryDispatchErrors: true,
+      suppressFailureFallback: true,
+    });
+
+    expect(result).toMatchObject({ kind: "failed-retryable" });
+    expect((result as { error?: unknown }).error).toBeInstanceOf(Error);
+    expect(deliverReplies).not.toHaveBeenCalled();
+  });
+
+  it("does not return retryable after spooled replay already showed visible output", async () => {
+    const { answerDraftStream } = setupDraftStreams({ answerMessageId: 2001 });
+    dispatchReplyWithBufferedBlockDispatcher.mockImplementation(async ({ dispatcherOptions }) => {
+      await dispatcherOptions.deliver({ text: "partial answer" }, { kind: "block" });
+      dispatcherOptions.onSkip?.({ text: "final answer" }, { kind: "final", reason: "empty" });
+      return { queuedFinal: false };
+    });
+
+    const result = await dispatchWithContext({
+      context: createContext(),
+      retryDispatchErrors: true,
+      suppressFailureFallback: true,
+    });
+
+    expect(result).toEqual({ kind: "completed" });
+    expect(answerDraftStream.update).toHaveBeenCalledWith("partial answer");
+    expect(deliverReplies).not.toHaveBeenCalled();
+  });
+
   it("keeps tool progress visible after a partial-streamed intermediate block", async () => {
     const { answerDraftStream } = setupDraftStreams({ answerMessageId: 2001 });
     dispatchReplyWithBufferedBlockDispatcher.mockImplementation(

extensions/telegram/src/bot-message-dispatch.ts

@@ -251,8 +251,14 @@ type DispatchTelegramMessageParams = {
   telegramCfg: TelegramAccountConfig;
   telegramDeps?: TelegramBotDeps;
   opts: Pick<TelegramBotOptions, "token" | "mediaMaxMb">;
+  retryDispatchErrors?: boolean;
+  suppressFailureFallback?: boolean;
 };
 
+export type TelegramDispatchResult =
+  | { kind: "completed" }
+  | { kind: "failed-retryable"; error: unknown };
+
 type TelegramReasoningLevel = "off" | "on" | "stream";
 
 type TelegramTranscriptMirrorPayload = { text?: string; mediaUrls?: string[] };
@@ -721,7 +727,9 @@ export const dispatchTelegramMessage = async ({
   telegramCfg,
   telegramDeps: injectedTelegramDeps,
   opts,
-}: DispatchTelegramMessageParams) => {
+  retryDispatchErrors = false,
+  suppressFailureFallback = false,
+}: DispatchTelegramMessageParams): Promise<TelegramDispatchResult> => {
   const dispatchStartedAt = Date.now();
   const dispatchContext = resolveDispatchTelegramContext({ cfg, context });
   const telegramDeps =
@@ -2471,7 +2479,7 @@ export const dispatchTelegramMessage = async ({
         },
       });
       if (!turnResult.dispatched) {
-        return;
+        return { kind: "completed" };
       }
       ({ queuedFinal } = turnResult.dispatchResult);
       suppressSilentReplyFallback =
@@ -2539,12 +2547,13 @@ export const dispatchTelegramMessage = async ({
     if (!isRoomEvent || deliveryState.snapshot().delivered) {
       clearGroupHistory();
     }
-    return;
+    return { kind: "completed" };
   }
   let sentFallback = false;
   const deliverySummary = deliveryState.snapshot();
   const shouldSendFailureFallback =
     !isRoomEvent &&
+    !suppressFailureFallback &&
     (dispatchError ||
       (!deliverySummary.delivered &&
         (deliverySummary.skippedNonSilent > 0 || deliverySummary.failedNonSilent > 0)));
@@ -2599,6 +2608,16 @@ export const dispatchTelegramMessage = async ({
 
   const hasFinalResponse =
     deliverySummary.delivered || sentFallback || suppressSilentReplyFallback || queuedFinal;
+  const deliveryFailureWithoutFinalResponse =
+    !deliverySummary.delivered &&
+    (deliverySummary.skippedNonSilent > 0 || deliverySummary.failedNonSilent > 0);
+  const retryableDispatchFailure =
+    dispatchError ??
+    (deliveryFailureWithoutFinalResponse
+      ? new Error(
+          `Telegram reply delivery failed without a final response (failed=${deliverySummary.failedNonSilent}, skipped=${deliverySummary.skippedNonSilent})`,
+        )
+      : null);
 
   if (statusReactionController && !hasFinalResponse) {
     void finalizeTelegramStatusReaction({ outcome: "error", hasFinalResponse: false }).catch(
@@ -2611,12 +2630,16 @@ export const dispatchTelegramMessage = async ({
   const shouldClearGroupHistory =
     !isRoomEvent || deliverySummary.delivered || sentFallback || queuedFinal;
 
+  if (retryableDispatchFailure && retryDispatchErrors && !hasFinalResponse) {
+    return { kind: "failed-retryable", error: retryableDispatchFailure };
+  }
+
   if (!hasFinalResponse) {
     if (!shouldClearGroupHistory) {
-      return;
+      return { kind: "completed" };
     }
     clearGroupHistory();
-    return;
+    return { kind: "completed" };
   }
 
   // Fire-and-forget: auto-rename DM topic on first message.
@@ -2690,4 +2713,5 @@ export const dispatchTelegramMessage = async ({
   if (shouldClearGroupHistory) {
     clearGroupHistory();
   }
+  return { kind: "completed" };
 };

extensions/telegram/src/bot-message.test.ts

@@ -30,11 +30,15 @@ vi.mock("./bot-message-dispatch.js", () => ({
 
 let createTelegramMessageProcessor: typeof import("./bot-message.js").createTelegramMessageProcessor;
 let formatTelegramInboundLogLine: typeof import("./bot-message.js").formatTelegramInboundLogLine;
+let runWithTelegramUpdateProcessingFrame: typeof import("./bot-processing-outcome.js").runWithTelegramUpdateProcessingFrame;
+let withTelegramSpooledReplayUpdate: typeof import("./bot-processing-outcome.js").withTelegramSpooledReplayUpdate;
 
 describe("telegram bot message processor", () => {
   beforeAll(async () => {
     ({ createTelegramMessageProcessor, formatTelegramInboundLogLine } =
       await import("./bot-message.js"));
+    ({ runWithTelegramUpdateProcessingFrame, withTelegramSpooledReplayUpdate } =
+      await import("./bot-processing-outcome.js"));
   });
 
   beforeEach(() => {
@@ -74,6 +78,8 @@ describe("telegram bot message processor", () => {
   async function processSampleMessage(
     processMessage: ReturnType<typeof createTelegramMessageProcessor>,
     lifecycle?: import("./bot-message.js").TelegramMessageProcessorLifecycle,
+    primaryCtxOverrides: Record<string, unknown> = {},
+    options: Parameters<typeof processMessage>[3] = {},
   ) {
     return await processMessage(
       {
@@ -81,10 +87,11 @@ describe("telegram bot message processor", () => {
           chat: { id: 123, type: "private", title: "chat" },
           message_id: 456,
         },
+        ...primaryCtxOverrides,
       } as unknown as Parameters<typeof processMessage>[0],
       [],
       [],
-      {},
+      options,
       undefined,
       undefined,
       undefined,
@@ -97,14 +104,15 @@ describe("telegram bot message processor", () => {
     sendMessage: ReturnType<typeof vi.fn>,
   ) {
     const runtimeError = vi.fn();
+    const dispatchError = new Error("dispatch exploded");
     buildTelegramMessageContext.mockResolvedValue(createMessageContext(context));
-    dispatchTelegramMessage.mockRejectedValue(new Error("dispatch exploded"));
+    dispatchTelegramMessage.mockRejectedValue(dispatchError);
     const processMessage = createTelegramMessageProcessor({
       ...baseDeps,
       bot: { api: { sendMessage } },
       runtime: { error: runtimeError },
     } as unknown as Parameters<typeof createTelegramMessageProcessor>[0]);
-    return { processMessage, runtimeError };
+    return { processMessage, runtimeError, dispatchError };
   }
 
   function createMessageContext(context: Record<string, unknown> = {}) {
@@ -132,7 +140,7 @@ describe("telegram bot message processor", () => {
     );
 
     const processMessage = createTelegramMessageProcessor(baseDeps);
-    await expect(processSampleMessage(processMessage)).resolves.toBe(true);
+    await expect(processSampleMessage(processMessage)).resolves.toEqual({ kind: "completed" });
 
     expect(sendTyping).toHaveBeenCalledTimes(1);
     expect(dispatchTelegramMessage).toHaveBeenCalledTimes(1);
@@ -154,7 +162,9 @@ describe("telegram bot message processor", () => {
     );
 
     const processMessage = createTelegramMessageProcessor(baseDeps);
-    await expect(processSampleMessage(processMessage, { onDispatchStart })).resolves.toBe(true);
+    await expect(processSampleMessage(processMessage, { onDispatchStart })).resolves.toEqual({
+      kind: "completed",
+    });
 
     expect(sendTyping).toHaveBeenCalledTimes(1);
     expect(onDispatchStart).toHaveBeenCalledTimes(1);
@@ -172,7 +182,9 @@ describe("telegram bot message processor", () => {
     buildTelegramMessageContext.mockResolvedValue(null);
 
     const processMessage = createTelegramMessageProcessor(baseDeps);
-    await expect(processSampleMessage(processMessage, { onDispatchStart })).resolves.toBe(false);
+    await expect(processSampleMessage(processMessage, { onDispatchStart })).resolves.toEqual({
+      kind: "skipped",
+    });
 
     expect(onDispatchStart).not.toHaveBeenCalled();
     expect(dispatchTelegramMessage).not.toHaveBeenCalled();
@@ -194,7 +206,7 @@ describe("telegram bot message processor", () => {
     );
 
     const processMessage = createTelegramMessageProcessor(baseDeps);
-    await expect(processSampleMessage(processMessage)).resolves.toBe(true);
+    await expect(processSampleMessage(processMessage)).resolves.toEqual({ kind: "completed" });
 
     expect(sendTyping).not.toHaveBeenCalled();
     expect(dispatchTelegramMessage).toHaveBeenCalledTimes(1);
@@ -203,7 +215,7 @@ describe("telegram bot message processor", () => {
   it("skips dispatch when no context is produced", async () => {
     buildTelegramMessageContext.mockResolvedValue(null);
     const processMessage = createTelegramMessageProcessor(baseDeps);
-    await expect(processSampleMessage(processMessage)).resolves.toBe(false);
+    await expect(processSampleMessage(processMessage)).resolves.toEqual({ kind: "skipped" });
     expect(dispatchTelegramMessage).not.toHaveBeenCalled();
     expect(telegramInboundInfo).not.toHaveBeenCalled();
   });
@@ -237,7 +249,7 @@ describe("telegram bot message processor", () => {
     );
 
     const processMessage = createTelegramMessageProcessor(baseDeps);
-    await expect(processSampleMessage(processMessage)).resolves.toBe(true);
+    await expect(processSampleMessage(processMessage)).resolves.toEqual({ kind: "completed" });
 
     expect(sendTyping).toHaveBeenCalledTimes(1);
     expect(dispatchTelegramMessage).toHaveBeenCalledTimes(1);
@@ -245,7 +257,7 @@ describe("telegram bot message processor", () => {
 
   it("sends user-visible fallback when dispatch throws", async () => {
     const sendMessage = vi.fn().mockResolvedValue(undefined);
-    const { processMessage, runtimeError } = createDispatchFailureHarness(
+    const { processMessage, runtimeError, dispatchError } = createDispatchFailureHarness(
       {
         chatId: 123,
         threadSpec: { id: 456, scope: "forum" },
@@ -253,7 +265,9 @@ describe("telegram bot message processor", () => {
       },
       sendMessage,
     );
-    await expect(processSampleMessage(processMessage)).resolves.toBe(true);
+    const result = await processSampleMessage(processMessage);
+
+    expect(result).toEqual({ kind: "failed-retryable", error: dispatchError });
 
     expect(sendMessage).toHaveBeenCalledWith(
       123,
@@ -265,9 +279,104 @@ describe("telegram bot message processor", () => {
     );
   });
 
+  it("suppresses user-visible fallback while replaying a spooled update", async () => {
+    const sendMessage = vi.fn().mockResolvedValue(undefined);
+    const { processMessage, runtimeError, dispatchError } = createDispatchFailureHarness(
+      {
+        chatId: 123,
+        route: { sessionKey: "agent:main:main" },
+      },
+      sendMessage,
+    );
+    const update = { update_id: 123456 };
+    const result = await withTelegramSpooledReplayUpdate(update, async () =>
+      processSampleMessage(processMessage, undefined, { update }),
+    );
+
+    expect(result).toEqual({ kind: "failed-retryable", error: dispatchError });
+    expect(sendMessage).not.toHaveBeenCalled();
+    expect(runtimeError).toHaveBeenCalledWith(
+      "telegram message processing failed: Error: dispatch exploded",
+    );
+  });
+
+  it("suppresses user-visible fallback for synthetic buffered spooled replay contexts", async () => {
+    const sendMessage = vi.fn().mockResolvedValue(undefined);
+    const { processMessage, runtimeError, dispatchError } = createDispatchFailureHarness(
+      {
+        chatId: 123,
+        route: { sessionKey: "agent:main:main" },
+      },
+      sendMessage,
+    );
+    const result = await processSampleMessage(
+      processMessage,
+      undefined,
+      {},
+      { spooledReplay: true },
+    );
+
+    expect(result).toEqual({ kind: "failed-retryable", error: dispatchError });
+    expect(sendMessage).not.toHaveBeenCalled();
+    expect(dispatchTelegramMessage).toHaveBeenCalledWith(
+      expect.objectContaining({
+        retryDispatchErrors: true,
+        suppressFailureFallback: true,
+      }),
+    );
+    expect(runtimeError).toHaveBeenCalledWith(
+      "telegram message processing failed: Error: dispatch exploded",
+    );
+  });
+
+  it("does not record buffered spooled replay failures into the ambient update frame", async () => {
+    const sendMessage = vi.fn().mockResolvedValue(undefined);
+    const { processMessage, dispatchError } = createDispatchFailureHarness(
+      {
+        chatId: 123,
+        route: { sessionKey: "agent:main:main" },
+      },
+      sendMessage,
+    );
+
+    const frame = await runWithTelegramUpdateProcessingFrame(async () =>
+      processSampleMessage(processMessage, undefined, {}, { spooledReplay: true }),
+    );
+
+    expect(frame.value).toEqual({ kind: "failed-retryable", error: dispatchError });
+    expect(frame.result).toBeUndefined();
+  });
+
+  it("propagates spooled dispatcher failure results without sending fallback", async () => {
+    const sendMessage = vi.fn().mockResolvedValue(undefined);
+    const dispatchError = new Error("agent dispatch failed");
+    const runtimeError = vi.fn();
+    buildTelegramMessageContext.mockResolvedValue(createMessageContext({ chatId: 123 }));
+    dispatchTelegramMessage.mockResolvedValue({ kind: "failed-retryable", error: dispatchError });
+    const processMessage = createTelegramMessageProcessor({
+      ...baseDeps,
+      bot: { api: { sendMessage } },
+      runtime: { error: runtimeError },
+    } as unknown as Parameters<typeof createTelegramMessageProcessor>[0]);
+    const update = { update_id: 123457 };
+    const result = await withTelegramSpooledReplayUpdate(update, async () =>
+      processSampleMessage(processMessage, undefined, { update }),
+    );
+
+    expect(result).toEqual({ kind: "failed-retryable", error: dispatchError });
+    expect(sendMessage).not.toHaveBeenCalled();
+    expect(dispatchTelegramMessage).toHaveBeenCalledWith(
+      expect.objectContaining({
+        retryDispatchErrors: true,
+        suppressFailureFallback: true,
+      }),
+    );
+    expect(runtimeError).not.toHaveBeenCalled();
+  });
+
   it("omits message_thread_id for General-topic fallback replies", async () => {
     const sendMessage = vi.fn().mockResolvedValue(undefined);
-    const { processMessage } = createDispatchFailureHarness(
+    const { processMessage, dispatchError } = createDispatchFailureHarness(
       {
         chatId: 123,
         threadSpec: { id: 1, scope: "forum" },
@@ -275,7 +384,9 @@ describe("telegram bot message processor", () => {
       },
       sendMessage,
     );
-    await expect(processSampleMessage(processMessage)).resolves.toBe(true);
+    const result = await processSampleMessage(processMessage);
+
+    expect(result).toEqual({ kind: "failed-retryable", error: dispatchError });
 
     expect(sendMessage).toHaveBeenCalledWith(
       123,
@@ -286,14 +397,16 @@ describe("telegram bot message processor", () => {
 
   it("swallows fallback delivery failures after dispatch throws", async () => {
     const sendMessage = vi.fn().mockRejectedValue(new Error("blocked by user"));
-    const { processMessage, runtimeError } = createDispatchFailureHarness(
+    const { processMessage, runtimeError, dispatchError } = createDispatchFailureHarness(
       {
         chatId: 123,
         route: { sessionKey: "agent:main:main" },
       },
       sendMessage,
     );
-    await expect(processSampleMessage(processMessage)).resolves.toBe(true);
+    const result = await processSampleMessage(processMessage);
+
+    expect(result).toEqual({ kind: "failed-retryable", error: dispatchError });
 
     expect(sendMessage).toHaveBeenCalledWith(
       123,

extensions/telegram/src/bot-message.ts

@@ -17,6 +17,11 @@ import {
 import type { TelegramMessageContextOptions } from "./bot-message-context.types.js";
 import type { TelegramPromptContextEntry } from "./bot-message-context.types.js";
 import { dispatchTelegramMessage } from "./bot-message-dispatch.js";
+import {
+  isTelegramSpooledReplayUpdate,
+  recordTelegramMessageProcessingResult,
+  type TelegramMessageProcessingResult,
+} from "./bot-processing-outcome.js";
 import type { TelegramBotOptions } from "./bot.types.js";
 import { buildTelegramThreadParams } from "./bot/helpers.js";
 import type { TelegramContext, TelegramStreamMode } from "./bot/types.js";
@@ -118,6 +123,12 @@ export const createTelegramMessageProcessor = (deps: TelegramMessageProcessorDep
     const ingressDebugEnabled =
       shouldLogVerbose() || process.env.OPENCLAW_DEBUG_TELEGRAM_INGRESS === "1";
     const ingressContextStartMs = ingressReceivedAtMs ? Date.now() : undefined;
+    const recordCurrentUpdateProcessingResult = (result: TelegramMessageProcessingResult) => {
+      if (options?.spooledReplay === true) {
+        return;
+      }
+      recordTelegramMessageProcessingResult(result);
+    };
     const context = await buildTelegramMessageContext({
       primaryCtx,
       allMedia,
@@ -152,7 +163,9 @@ export const createTelegramMessageProcessor = (deps: TelegramMessageProcessorDep
             (options?.ingressBuffer ? ` buffer=${options.ingressBuffer}` : ""),
         );
       }
-      return false;
+      const result: TelegramMessageProcessingResult = { kind: "skipped" };
+      recordCurrentUpdateProcessingResult(result);
+      return result;
     }
     if (ingressDebugEnabled && ingressReceivedAtMs && ingressContextStartMs) {
       logVerbose(
@@ -181,8 +194,10 @@ export const createTelegramMessageProcessor = (deps: TelegramMessageProcessorDep
       }),
     );
     await lifecycle?.onDispatchStart?.();
+    const spooledReplay =
+      options?.spooledReplay === true || isTelegramSpooledReplayUpdate(primaryCtx.update);
     try {
-      await dispatchTelegramMessage({
+      const dispatchResult = await dispatchTelegramMessage({
         context,
         bot,
         cfg,
@@ -193,23 +208,43 @@ export const createTelegramMessageProcessor = (deps: TelegramMessageProcessorDep
         telegramCfg,
         telegramDeps,
         opts,
+        retryDispatchErrors: spooledReplay,
+        suppressFailureFallback: spooledReplay,
       });
+      if (dispatchResult?.kind === "failed-retryable") {
+        const result: TelegramMessageProcessingResult = {
+          kind: "failed-retryable",
+          error: dispatchResult.error,
+        };
+        recordCurrentUpdateProcessingResult(result);
+        return result;
+      }
       if (ingressDebugEnabled && ingressReceivedAtMs) {
         logVerbose(
           `telegram ingress: chatId=${context.chatId} dispatchCompleteMs=${Date.now() - ingressReceivedAtMs}` +
             (options?.ingressBuffer ? ` buffer=${options.ingressBuffer}` : ""),
         );
       }
+      const result: TelegramMessageProcessingResult = { kind: "completed" };
+      recordCurrentUpdateProcessingResult(result);
+      return result;
     } catch (err) {
       runtime.error?.(danger(`telegram message processing failed: ${String(err)}`));
-      try {
-        await bot.api.sendMessage(
-          context.chatId,
-          "Something went wrong while processing your request. Please try again.",
-          buildTelegramThreadParams(context.threadSpec),
-        );
-      } catch {}
+      if (!spooledReplay) {
+        try {
+          await bot.api.sendMessage(
+            context.chatId,
+            "Something went wrong while processing your request. Please try again.",
+            buildTelegramThreadParams(context.threadSpec),
+          );
+        } catch {}
+      }
+      const result: TelegramMessageProcessingResult = {
+        kind: "failed-retryable",
+        error: err,
+      };
+      recordCurrentUpdateProcessingResult(result);
+      return result;
     }
-    return true;
   };
 };

extensions/telegram/src/bot-native-commands.ts

@@ -65,6 +65,7 @@ import {
   syncTelegramMenuCommands as syncTelegramMenuCommandsRuntime,
   type TelegramMenuCommand,
 } from "./bot-native-command-menu.js";
+import type { TelegramMessageProcessingResult } from "./bot-processing-outcome.js";
 import type { TelegramUpdateKeyContext } from "./bot-updates.js";
 import type { TelegramBotOptions } from "./bot.types.js";
 import {
@@ -444,7 +445,7 @@ export type RegisterTelegramHandlerParams = {
     replyChain?: import("./message-cache.js").TelegramReplyChainEntry[],
     promptContext?: import("./bot-message-context.types.js").TelegramPromptContextEntry[],
     lifecycle?: import("./bot-message.js").TelegramMessageProcessorLifecycle,
-  ) => Promise<boolean>;
+  ) => Promise<TelegramMessageProcessingResult>;
   logger: ReturnType<typeof getChildLogger>;
 };
 

extensions/telegram/src/bot-processing-outcome.ts

@@ -0,0 +1,126 @@
+// Telegram plugin module tracks per-update processing outcomes.
+import { AsyncLocalStorage } from "node:async_hooks";
+
+export type TelegramMessageProcessingResult =
+  | { kind: "completed" }
+  | { kind: "skipped" }
+  | { kind: "failed-retryable"; error: unknown };
+
+type TelegramUpdateProcessingFrame = {
+  result?: TelegramMessageProcessingResult;
+};
+
+type TelegramSpooledReplayFrame = {
+  deferredWork?: TelegramSpooledReplayDeferredParticipant;
+};
+
+export type TelegramSpooledReplayDeferredParticipant = {
+  key: string;
+  task: Promise<TelegramMessageProcessingResult>;
+  settle: (result: TelegramMessageProcessingResult) => void;
+};
+
+const telegramUpdateProcessingFrames = new AsyncLocalStorage<TelegramUpdateProcessingFrame>();
+const telegramSpooledReplayFrames = new AsyncLocalStorage<TelegramSpooledReplayFrame>();
+const telegramSpooledReplayUpdates = new WeakSet<object>();
+
+export class TelegramSpooledReplayProcessingError extends Error {
+  override readonly cause: unknown;
+
+  constructor(cause: unknown) {
+    super(`telegram spooled update processing failed: ${String(cause)}`);
+    this.name = "TelegramSpooledReplayProcessingError";
+    this.cause = cause;
+  }
+}
+
+export async function runWithTelegramUpdateProcessingFrame<T>(
+  fn: () => Promise<T>,
+): Promise<{ value: T; result?: TelegramMessageProcessingResult }> {
+  const frame: TelegramUpdateProcessingFrame = {};
+  const value = await telegramUpdateProcessingFrames.run(frame, fn);
+  return frame.result ? { value, result: frame.result } : { value };
+}
+
+export function recordTelegramMessageProcessingResult(
+  result: TelegramMessageProcessingResult,
+): void {
+  const frame = telegramUpdateProcessingFrames.getStore();
+  if (!frame) {
+    return;
+  }
+  if (result.kind === "failed-retryable") {
+    frame.result = result;
+    return;
+  }
+  if (!frame.result || frame.result.kind === "skipped") {
+    frame.result = result;
+  }
+}
+
+function createTelegramSpooledReplayParticipant(
+  key: string,
+): TelegramSpooledReplayDeferredParticipant {
+  let settled = false;
+  let resolveTask: (result: TelegramMessageProcessingResult) => void = () => {};
+  const task = new Promise<TelegramMessageProcessingResult>((resolve) => {
+    resolveTask = resolve;
+  });
+  return {
+    key,
+    task,
+    settle: (result) => {
+      if (settled) {
+        return;
+      }
+      settled = true;
+      resolveTask(result);
+    },
+  };
+}
+
+export function createTelegramSpooledReplayDeferredParticipant(
+  key: string,
+): TelegramSpooledReplayDeferredParticipant | null {
+  const frame = telegramSpooledReplayFrames.getStore();
+  if (!frame) {
+    return null;
+  }
+  const participant = createTelegramSpooledReplayParticipant(key);
+  frame.deferredWork = participant;
+  return participant;
+}
+
+export function getTelegramSpooledReplayDeferredParticipant():
+  | TelegramSpooledReplayDeferredParticipant
+  | undefined {
+  return telegramSpooledReplayFrames.getStore()?.deferredWork;
+}
+
+export async function runWithTelegramSpooledReplayUpdate<T>(
+  update: object,
+  fn: () => Promise<T>,
+): Promise<{ value: T; deferredWork?: TelegramSpooledReplayDeferredParticipant }> {
+  const frame: TelegramSpooledReplayFrame = {};
+  telegramSpooledReplayUpdates.add(update);
+  try {
+    const value = await telegramSpooledReplayFrames.run(frame, fn);
+    return frame.deferredWork ? { value, deferredWork: frame.deferredWork } : { value };
+  } finally {
+    telegramSpooledReplayUpdates.delete(update);
+  }
+}
+
+export async function withTelegramSpooledReplayUpdate<T>(
+  update: object,
+  fn: () => Promise<T>,
+): Promise<T> {
+  return (await runWithTelegramSpooledReplayUpdate(update, fn)).value;
+}
+
+export function isTelegramSpooledReplayUpdate(update: unknown): boolean {
+  return (
+    telegramSpooledReplayFrames.getStore() !== undefined ||
+    (typeof update === "object" && update !== null && telegramSpooledReplayUpdates.has(update))
+  );
+}

extensions/telegram/src/bot.create-telegram-bot.test.ts

@@ -63,6 +63,13 @@ const {
   resolveTelegramScopedGroupConfig,
   setTelegramBotRuntimeForTest,
 } = await import("./bot-core.js");
+const {
+  createTelegramSpooledReplayDeferredParticipant,
+  recordTelegramMessageProcessingResult,
+  runWithTelegramSpooledReplayUpdate,
+  TelegramSpooledReplayProcessingError,
+  withTelegramSpooledReplayUpdate,
+} = await import("./bot-processing-outcome.js");
 const { resolveTelegramConversationRoute } = await import("./conversation-route.js");
 const { clearAccountThrottlersForTest } = await import("./account-throttler.js");
 const {
@@ -841,6 +848,146 @@ describe("createTelegramBot", () => {
     },
   );
 
+  it("settles spooled replay participants when stop cancels pending inbound debounce", async () => {
+    const DEBOUNCE_MS = 4321;
+    loadConfig.mockReturnValue({
+      agents: {
+        defaults: {
+          envelopeTimezone: "utc",
+        },
+      },
+      messages: {
+        inbound: {
+          debounceMs: DEBOUNCE_MS,
+        },
+      },
+      channels: {
+        telegram: { dmPolicy: "open", allowFrom: ["*"] },
+      },
+    });
+
+    installPerKeySequentializer();
+    const setTimeoutSpy = vi.spyOn(globalThis, "setTimeout");
+
+    try {
+      createTelegramBot({ token: "tok" });
+      const messageHandler = getOnHandler("message") as (
+        ctx: TelegramMiddlewareTestContext,
+      ) => Promise<void>;
+      const firstUpdate = { update_id: 201 };
+      const replay = await runWithTelegramSpooledReplayUpdate(firstUpdate, async () => {
+        await runTelegramMiddlewareChain({
+          ctx: {
+            update: firstUpdate,
+            message: {
+              chat: { id: 7, type: "private" },
+              text: "first",
+              date: 1736380800,
+              message_id: 201,
+              from: { id: 42, first_name: "Ada" },
+            },
+            me: { username: "openclaw_bot" },
+            getFile: async () => ({}),
+          },
+          finalHandler: messageHandler,
+        });
+      });
+
+      const deferredWork = replay.deferredWork;
+      expect(deferredWork).toBeDefined();
+      if (!deferredWork) {
+        throw new Error("Expected spooled replay deferred work");
+      }
+      await runTelegramMiddlewareChain({
+        ctx: {
+          update: { update_id: 202 },
+          message: {
+            chat: { id: 7, type: "private" },
+            text: "stop",
+            date: 1736380801,
+            message_id: 202,
+            from: { id: 42, first_name: "Ada" },
+          },
+          me: { username: "openclaw_bot" },
+          getFile: async () => ({}),
+        },
+        finalHandler: messageHandler,
+      });
+
+      await expect(deferredWork.task).resolves.toEqual({ kind: "skipped" });
+      const debounceCallIndex = setTimeoutSpy.mock.calls.findLastIndex(
+        (call) => call[1] === DEBOUNCE_MS,
+      );
+      expect(debounceCallIndex).toBeGreaterThanOrEqual(0);
+      clearTimeout(
+        setTimeoutSpy.mock.results[debounceCallIndex]?.value as ReturnType<typeof setTimeout>,
+      );
+    } finally {
+      setTimeoutSpy.mockRestore();
+    }
+  });
+
+  it("settles spooled replay participants when stop cancels pending text fragments", async () => {
+    loadConfig.mockReturnValue({
+      agents: {
+        defaults: {
+          envelopeTimezone: "utc",
+        },
+      },
+      channels: {
+        telegram: { dmPolicy: "open", allowFrom: ["*"] },
+      },
+    });
+
+    installPerKeySequentializer();
+
+    createTelegramBot({ token: "tok" });
+    const messageHandler = getOnHandler("message") as (
+      ctx: TelegramMiddlewareTestContext,
+    ) => Promise<void>;
+    const firstUpdate = { update_id: 211 };
+    const replay = await runWithTelegramSpooledReplayUpdate(firstUpdate, async () => {
+      await runTelegramMiddlewareChain({
+        ctx: {
+          update: firstUpdate,
+          message: {
+            chat: { id: 7, type: "private" },
+            text: "A".repeat(4050),
+            date: 1736380800,
+            message_id: 211,
+            from: { id: 42, first_name: "Ada" },
+          },
+          me: { username: "openclaw_bot" },
+          getFile: async () => ({}),
+        },
+        finalHandler: messageHandler,
+      });
+    });
+
+    const deferredWork = replay.deferredWork;
+    expect(deferredWork).toBeDefined();
+    if (!deferredWork) {
+      throw new Error("Expected spooled replay deferred work");
+    }
+    await runTelegramMiddlewareChain({
+      ctx: {
+        update: { update_id: 212 },
+        message: {
+          chat: { id: 7, type: "private" },
+          text: "stop",
+          date: 1736380801,
+          message_id: 212,
+          from: { id: 42, first_name: "Ada" },
+        },
+        me: { username: "openclaw_bot" },
+        getFile: async () => ({}),
+      },
+      finalHandler: messageHandler,
+    });
+
+    await expect(deferredWork.task).resolves.toEqual({ kind: "skipped" });
+  });
+
   it("lets stop cancel pending same-chat forwarded debounce", async () => {
     const DEBOUNCE_MS = 4321;
     loadConfig.mockReturnValue({
@@ -2349,6 +2496,213 @@ describe("createTelegramBot", () => {
     expect(onUpdateId.mock.calls.map((call) => Number(call[0]))).toEqual([202]);
   });
 
+  it("persists recorded dispatch failures during normal polling", async () => {
+    sequentializeSpy.mockImplementationOnce(
+      () => async (_ctx: unknown, next: () => Promise<void>) => {
+        await next();
+      },
+    );
+
+    const onUpdateId = vi.fn();
+
+    createTelegramBot({
+      token: "tok",
+      updateOffset: {
+        lastUpdateId: 500,
+        onUpdateId,
+      },
+    });
+
+    type Middleware = (
+      ctx: Record<string, unknown>,
+      next: () => Promise<void>,
+    ) => Promise<void> | void;
+
+    const middlewares = middlewareUseSpy.mock.calls
+      .map((call) => call[0])
+      .filter((fn): fn is Middleware => typeof fn === "function");
+
+    const runMiddlewareChain = async (
+      ctx: Record<string, unknown>,
+      finalNext: () => Promise<void>,
+    ) => {
+      let idx = -1;
+      const dispatch = async (i: number): Promise<void> => {
+        if (i <= idx) {
+          throw new Error("middleware dispatch called multiple times");
+        }
+        idx = i;
+        const fn = middlewares[i];
+        if (!fn) {
+          await finalNext();
+          return;
+        }
+        await fn(ctx, async () => dispatch(i + 1));
+      };
+      await dispatch(0);
+    };
+
+    const dispatchError = new Error("dispatch exploded");
+    await runMiddlewareChain({ update: { update_id: 501 } }, async () => {
+      recordTelegramMessageProcessingResult({
+        kind: "failed-retryable",
+        error: dispatchError,
+      });
+    });
+    await flushTelegramTestMicrotasks();
+    expect(onUpdateId.mock.calls.map((call) => Number(call[0]))).toEqual([501]);
+
+    await runMiddlewareChain({ update: { update_id: 502 } }, async () => {});
+    await flushTelegramTestMicrotasks();
+    expect(onUpdateId.mock.calls.map((call) => Number(call[0]))).toEqual([501, 502]);
+  });
+
+  it("rejects recorded dispatch failures during isolated spool replay", async () => {
+    sequentializeSpy.mockImplementationOnce(
+      () => async (_ctx: unknown, next: () => Promise<void>) => {
+        await next();
+      },
+    );
+
+    const onUpdateId = vi.fn();
+
+    createTelegramBot({
+      token: "tok",
+      updateOffset: {
+        lastUpdateId: 600,
+        onUpdateId,
+      },
+    });
+
+    type Middleware = (
+      ctx: Record<string, unknown>,
+      next: () => Promise<void>,
+    ) => Promise<void> | void;
+
+    const middlewares = middlewareUseSpy.mock.calls
+      .map((call) => call[0])
+      .filter((fn): fn is Middleware => typeof fn === "function");
+
+    const runMiddlewareChain = async (
+      ctx: Record<string, unknown>,
+      finalNext: () => Promise<void>,
+    ) => {
+      let idx = -1;
+      const dispatch = async (i: number): Promise<void> => {
+        if (i <= idx) {
+          throw new Error("middleware dispatch called multiple times");
+        }
+        idx = i;
+        const fn = middlewares[i];
+        if (!fn) {
+          await finalNext();
+          return;
+        }
+        await fn(ctx, async () => dispatch(i + 1));
+      };
+      await dispatch(0);
+    };
+
+    const update = { update_id: 601 };
+    const dispatchError = new Error("dispatch exploded");
+    await expect(
+      withTelegramSpooledReplayUpdate(update, async () => {
+        await runMiddlewareChain({ update }, async () => {
+          recordTelegramMessageProcessingResult({
+            kind: "failed-retryable",
+            error: dispatchError,
+          });
+        });
+      }),
+    ).rejects.toMatchObject({
+      name: TelegramSpooledReplayProcessingError.name,
+      cause: dispatchError,
+    });
+    await flushTelegramTestMicrotasks();
+    expect(onUpdateId).not.toHaveBeenCalled();
+  });
+
+  it("keeps deferred spooled failures retryable in the same bot tracker", async () => {
+    sequentializeSpy.mockImplementationOnce(
+      () => async (_ctx: unknown, next: () => Promise<void>) => {
+        await next();
+      },
+    );
+
+    const onUpdateId = vi.fn();
+
+    createTelegramBot({
+      token: "tok",
+      updateOffset: {
+        lastUpdateId: 700,
+        onUpdateId,
+      },
+    });
+
+    type Middleware = (
+      ctx: Record<string, unknown>,
+      next: () => Promise<void>,
+    ) => Promise<void> | void;
+
+    const middlewares = middlewareUseSpy.mock.calls
+      .map((call) => call[0])
+      .filter((fn): fn is Middleware => typeof fn === "function");
+
+    const runMiddlewareChain = async (
+      ctx: Record<string, unknown>,
+      finalNext: () => Promise<void>,
+    ) => {
+      let idx = -1;
+      const dispatch = async (i: number): Promise<void> => {
+        if (i <= idx) {
+          throw new Error("middleware dispatch called multiple times");
+        }
+        idx = i;
+        const fn = middlewares[i];
+        if (!fn) {
+          await finalNext();
+          return;
+        }
+        await fn(ctx, async () => dispatch(i + 1));
+      };
+      await dispatch(0);
+    };
+
+    const update = { update_id: 701 };
+    const replay = await runWithTelegramSpooledReplayUpdate(update, async () => {
+      await runMiddlewareChain({ update }, async () => {
+        const participant = createTelegramSpooledReplayDeferredParticipant("test:deferred");
+        if (!participant) {
+          throw new Error("expected spooled replay participant");
+        }
+      });
+    });
+    const deferredWork = replay.deferredWork;
+    expect(deferredWork).toBeDefined();
+    if (!deferredWork) {
+      throw new Error("Expected deferred spooled work");
+    }
+    await flushTelegramTestMicrotasks();
+    expect(onUpdateId).not.toHaveBeenCalled();
+
+    deferredWork.settle({
+      kind: "failed-retryable",
+      error: new Error("deferred dispatch failed"),
+    });
+    await flushTelegramTestMicrotasks();
+    expect(onUpdateId).not.toHaveBeenCalled();
+
+    let retried = false;
+    await runWithTelegramSpooledReplayUpdate(update, async () => {
+      await runMiddlewareChain({ update }, async () => {
+        retried = true;
+      });
+    });
+    await flushTelegramTestMicrotasks();
+    expect(retried).toBe(true);
+    expect(onUpdateId.mock.calls.map((call) => Number(call[0]))).toEqual([701]);
+  });
+
   it("skips replayed update ids even when the semantic update key differs", async () => {
     sequentializeSpy.mockImplementationOnce(
       () => async (_ctx: unknown, next: () => Promise<void>) => {

extensions/telegram/src/bot/types.ts

@@ -19,6 +19,7 @@ export type TelegramGetChat = (chatId: number | string) => Promise<TelegramChatD
  */
 export type TelegramContext = {
   message: Message;
+  update?: unknown;
   me?: UserFromGetMe;
   getFile: TelegramGetFile;
 };

extensions/telegram/src/draft-stream.test.ts

@@ -378,11 +378,20 @@ describe("createTelegramDraftStream", () => {
     expect(stream.sendMayHaveLanded?.()).toBe(expected);
   }
 
-  it("clears sendMayHaveLanded on pre-connect first preview send failures", async () => {
-    await expectSendMayHaveLandedStateAfterFirstFailure(
+  it("retries pre-connect first preview send failures instead of stopping", async () => {
+    const api = createMockDraftApi();
+    api.sendMessage.mockRejectedValueOnce(
       Object.assign(new Error("connect ECONNREFUSED"), { code: "ECONNREFUSED" }),
-      false,
     );
+    const stream = createDraftStream(api);
+
+    stream.update("Hello");
+    await stream.flush();
+    await stream.flush();
+
+    expect(api.sendMessage).toHaveBeenCalledTimes(2);
+    expect(stream.sendMayHaveLanded?.()).toBe(false);
+    expect(stream.messageId()).toBe(17);
   });
 
   it("clears sendMayHaveLanded on Telegram 4xx client rejections", async () => {
@@ -392,6 +401,103 @@ describe("createTelegramDraftStream", () => {
     );
   });
 
+  it("treats message-is-not-modified edits as delivered", async () => {
+    const api = createMockDraftApi();
+    api.editMessageText.mockRejectedValueOnce(
+      Object.assign(
+        new Error("Call to 'editMessageText' failed! (400: Bad Request: message is not modified)"),
+        { error_code: 400 },
+      ),
+    );
+    const warn = vi.fn();
+    const stream = createDraftStream(api, { warn });
+
+    stream.update("Hello");
+    await stream.flush();
+    stream.update("Hello again");
+    await stream.flush();
+    stream.update("Hello more");
+    await stream.flush();
+
+    expect(api.editMessageText).toHaveBeenCalledTimes(2);
+    expect(api.editMessageText).toHaveBeenLastCalledWith(123, 17, "Hello more");
+    expect(warn).not.toHaveBeenCalled();
+  });
+
+  it("retries the preview edit after a transient network failure", async () => {
+    const api = createMockDraftApi();
+    api.editMessageText.mockRejectedValueOnce(
+      Object.assign(new Error("read ECONNRESET"), { code: "ECONNRESET" }),
+    );
+    const warn = vi.fn();
+    const stream = createDraftStream(api, { warn });
+
+    stream.update("Hello");
+    await stream.flush();
+    stream.update("Hello again");
+    await stream.flush();
+    expect(warn).toHaveBeenCalledWith(
+      "telegram stream preview edit failed (retrying): read ECONNRESET",
+    );
+
+    await stream.flush();
+
+    expect(api.editMessageText).toHaveBeenCalledTimes(2);
+    expect(api.editMessageText).toHaveBeenLastCalledWith(123, 17, "Hello again");
+    expect(stream.lastDeliveredText?.()).toBe("Hello again");
+  });
+
+  it("suspends preview edits for retry_after during flood control", async () => {
+    vi.useFakeTimers();
+    try {
+      const api = createMockDraftApi();
+      api.editMessageText.mockRejectedValueOnce(
+        Object.assign(
+          new Error("Call to 'editMessageText' failed! (429: Too Many Requests: retry after 1)"),
+          { error_code: 429, parameters: { retry_after: 1 } },
+        ),
+      );
+      const stream = createDraftStream(api);
+
+      stream.update("Hello");
+      await stream.flush();
+      stream.update("Hello again");
+      await stream.flush();
+      stream.update("Hello more");
+      await stream.flush();
+      expect(api.editMessageText).toHaveBeenCalledTimes(1);
+
+      await vi.advanceTimersByTimeAsync(1100);
+      await stream.flush();
+
+      expect(api.editMessageText).toHaveBeenCalledTimes(2);
+      expect(api.editMessageText).toHaveBeenLastCalledWith(123, 17, "Hello more");
+    } finally {
+      vi.useRealTimers();
+    }
+  });
+
+  it("stops the preview after repeated retryable edit failures", async () => {
+    const api = createMockDraftApi();
+    api.editMessageText.mockRejectedValue(
+      Object.assign(new Error("read ECONNRESET"), { code: "ECONNRESET" }),
+    );
+    const warn = vi.fn();
+    const stream = createDraftStream(api, { warn });
+
+    stream.update("Hello");
+    await stream.flush();
+    stream.update("Hello again");
+    await stream.flush();
+    await stream.flush();
+    await stream.flush();
+    await stream.flush();
+    await stream.flush();
+
+    expect(api.editMessageText).toHaveBeenCalledTimes(4);
+    expect(warn).toHaveBeenCalledWith("telegram stream preview failed: read ECONNRESET");
+  });
+
   it("supports rendered previews with parse_mode", async () => {
     const api = createMockDraftApi();
     const stream = createTelegramDraftStream({

extensions/telegram/src/draft-stream.ts

@@ -6,11 +6,25 @@ import {
 } from "openclaw/plugin-sdk/channel-outbound";
 import { formatErrorMessage } from "openclaw/plugin-sdk/error-runtime";
 import { buildTelegramThreadParams, type TelegramThreadSpec } from "./bot/helpers.js";
-import { isSafeToRetrySendError, isTelegramClientRejection } from "./network-errors.js";
+import {
+  isRecoverableTelegramNetworkError,
+  isSafeToRetrySendError,
+  isTelegramClientRejection,
+  isTelegramMessageNotModifiedError,
+  isTelegramRateLimitError,
+  readTelegramRetryAfterMs,
+} from "./network-errors.js";
 import { normalizeTelegramReplyToMessageId } from "./outbound-params.js";
 
 const TELEGRAM_STREAM_MAX_CHARS = 4096;
 const DEFAULT_THROTTLE_MS = 1000;
+// Retryable preview failures keep the latest text pending for the next throttle
+// tick; cap consecutive misses so a persistent outage stops the preview instead
+// of warn-spamming for the rest of the run.
+const MAX_CONSECUTIVE_PREVIEW_FAILURES = 3;
+// Flood waits beyond this freeze the preview longer than it is useful; clamp so
+// a large retry_after cannot park the suspension past the run's lifetime.
+const MAX_PREVIEW_FLOOD_SUSPEND_MS = 60_000;
 
 export type TelegramDraftStream = {
   update: (text: string) => void;
@@ -109,6 +123,8 @@ export function createTelegramDraftStream(params: {
 
   const streamState = { stopped: false, final: false };
   let messageSendAttempted = false;
+  let suspendedUntilMs = 0;
+  let consecutivePreviewFailures = 0;
   let streamMessageId: number | undefined;
   let streamVisibleSinceMs: number | undefined;
   let lastSentText = "";
@@ -198,6 +214,12 @@ export function createTelegramDraftStream(params: {
     if (streamState.stopped && !streamState.final) {
       return false;
     }
+    // Flood-control suspension: returning false keeps the newest text pending,
+    // so the first tick after retry_after delivers it. Final flushes still try
+    // so the last text has a chance to land.
+    if (!streamState.final && Date.now() < suspendedUntilMs) {
+      return false;
+    }
     const trimmed = text.trimEnd();
     if (!trimmed) {
       return false;
@@ -262,6 +284,8 @@ export function createTelegramDraftStream(params: {
       }
     }
 
+    const previousSentText = lastSentText;
+    const previousSentParseMode = lastSentParseMode;
     lastSentText = renderedText;
     lastSentParseMode = renderedParseMode;
     try {
@@ -273,9 +297,40 @@ export function createTelegramDraftStream(params: {
       if (sent) {
         previewRevision += 1;
         lastDeliveredText = trimmed;
+        consecutivePreviewFailures = 0;
+        suspendedUntilMs = 0;
       }
       return sent;
     } catch (err) {
+      const isEdit = typeof streamMessageId === "number";
+      if (isEdit && isTelegramMessageNotModifiedError(err)) {
+        // Telegram already shows exactly this text; count the edit as delivered.
+        consecutivePreviewFailures = 0;
+        lastDeliveredText = trimmed;
+        return true;
+      }
+      // Roll back the dedupe snapshot so the retried tick is not skipped as a no-op.
+      lastSentText = previousSentText;
+      lastSentParseMode = previousSentParseMode;
+      // Flood control is always retryable: Telegram rejected the call outright.
+      // Beyond that, edits retry on any transient network error (re-editing the
+      // same content is idempotent) while an unsent first preview retries only
+      // on provably pre-connect failures — anything ambiguous could duplicate
+      // the preview message.
+      const retryable =
+        isTelegramRateLimitError(err) ||
+        (isEdit ? isRecoverableTelegramNetworkError(err) : isSafeToRetrySendError(err));
+      consecutivePreviewFailures += 1;
+      if (retryable && consecutivePreviewFailures <= MAX_CONSECUTIVE_PREVIEW_FAILURES) {
+        const retryAfterMs = readTelegramRetryAfterMs(err);
+        if (retryAfterMs !== undefined) {
+          suspendedUntilMs = Date.now() + Math.min(retryAfterMs, MAX_PREVIEW_FLOOD_SUSPEND_MS);
+        }
+        params.warn?.(
+          `telegram stream preview ${isEdit ? "edit" : "send"} failed (retrying): ${formatErrorMessage(err)}`,
+        );
+        return false;
+      }
       streamState.stopped = true;
       params.warn?.(`telegram stream preview failed: ${formatErrorMessage(err)}`);
       return false;

extensions/telegram/src/format.test.ts

@@ -55,6 +55,13 @@ describe("markdownToTelegramHtml", () => {
     ).toBe(input);
   });
 
+  it("preserves Telegram expandable blockquote HTML", () => {
+    const input = "<blockquote expandable>hidden details</blockquote>";
+
+    expect(markdownToTelegramHtml(input)).toBe(input);
+    expect(renderTelegramHtmlText(input, { textMode: "html" })).toBe(input);
+  });
+
   it("does not promote Telegram HTML tags inside code", () => {
     expect(markdownToTelegramHtml("`<b>literal</b>`")).toBe(
       "<code>&lt;b&gt;literal&lt;/b&gt;</code>",
@@ -66,6 +73,9 @@ describe("markdownToTelegramHtml", () => {
 
   it("keeps unsupported Telegram HTML variants escaped", () => {
     expect(markdownToTelegramHtml('<b class="x">bad</b>')).toBe('&lt;b class="x"&gt;bad&lt;/b&gt;');
+    expect(markdownToTelegramHtml('<blockquote cite="x">bad</blockquote>')).toBe(
+      '&lt;blockquote cite="x"&gt;bad&lt;/blockquote&gt;',
+    );
     expect(renderTelegramHtmlText('<b class="x">bad</b>', { textMode: "html" })).toBe(
       '&lt;b class="x"&gt;bad&lt;/b&gt;',
     );

extensions/telegram/src/format.ts

@@ -191,13 +191,13 @@ const TELEGRAM_SIMPLE_HTML_TAGS = new Set([
   "code",
   "pre",
   "tg-spoiler",
-  "blockquote",
 ]);
 const TELEGRAM_ATTR_HTML_TAG_PATTERNS = new Map([
   ["a", /^\s+href="[^"]+"\s*$/],
   ["span", /^\s+class="tg-spoiler"\s*$/],
   ["tg-emoji", /^\s+emoji-id="[^"]+"\s*$/],
   ["tg-time", /^\s+datetime="[^"]+"\s*$/],
+  ["blockquote", /^(\s+expandable)?\s*$/],
 ]);
 const TELEGRAM_CODE_LANGUAGE_ATTR_PATTERN = /^\s+class="language-[^"]+"\s*$/;
 let fileReferencePattern: RegExp | undefined;

extensions/telegram/src/message-dispatch-dedupe.test.ts

@@ -11,8 +11,11 @@ import {
   claimTelegramMessageDispatchReplay,
   commitTelegramMessageDispatchReplay,
   createTelegramMessageDispatchReplayGuard,
+  forgetTelegramMessageDispatchReplay,
   releaseTelegramMessageDispatchReplay,
   TELEGRAM_MESSAGE_DISPATCH_DEDUPE_NAMESPACE,
+  TelegramMessageDispatchReplayForgetError,
+  type TelegramMessageDispatchReplayGuard,
 } from "./message-dispatch-dedupe.js";
 
 const tempDirs: string[] = [];
@@ -204,4 +207,27 @@ describe("Telegram message dispatch replay guard", () => {
       key: first.key,
     });
   });
+
+  it("fails rollback when a committed dispatch key cannot be forgotten", async () => {
+    const guard = {
+      claim: async () => ({ kind: "claimed" }),
+      commit: async () => true,
+      forget: async (key: string) => key !== "failed-key",
+      hasRecent: async () => false,
+      warmup: async () => 0,
+      clearMemory: () => {},
+      memorySize: () => 0,
+      release: () => {},
+    } satisfies TelegramMessageDispatchReplayGuard;
+
+    await expect(
+      forgetTelegramMessageDispatchReplay({
+        guard,
+        keys: ["ok-key", "failed-key", "failed-key"],
+      }),
+    ).rejects.toMatchObject({
+      name: TelegramMessageDispatchReplayForgetError.name,
+      failures: [{ key: "failed-key" }],
+    });
+  });
 });

extensions/telegram/src/message-dispatch-dedupe.ts

@@ -11,13 +11,40 @@ export const TELEGRAM_MESSAGE_DISPATCH_DEDUPE_STATE_PLUGIN_ID = "telegram-messag
 export const TELEGRAM_MESSAGE_DISPATCH_DEDUPE_MEMORY_MAX_ENTRIES = 50_000;
 export const TELEGRAM_MESSAGE_DISPATCH_DEDUPE_STATE_MAX_ENTRIES = 50_000;
 
-export type TelegramMessageDispatchReplayGuard = ClaimableDedupe;
+export type TelegramMessageDispatchReplayGuard = ClaimableDedupe &
+  Required<Pick<ClaimableDedupe, "forget">>;
 
 export type TelegramMessageDispatchClaim =
   | { kind: "claimed"; key: string }
   | { kind: "duplicate" }
   | { kind: "invalid" };
 
+export type TelegramMessageDispatchReplayForgetFailure = {
+  key: string;
+  error?: unknown;
+};
+
+export class TelegramMessageDispatchReplayForgetError extends Error {
+  readonly failures: TelegramMessageDispatchReplayForgetFailure[];
+  override readonly cause: unknown;
+
+  constructor(failures: readonly TelegramMessageDispatchReplayForgetFailure[]) {
+    const count = failures.length;
+    super(`telegram message dispatch dedupe rollback failed for ${count} key(s)`, {
+      cause: failures.find((failure) => failure.error !== undefined)?.error,
+    });
+    this.name = "TelegramMessageDispatchReplayForgetError";
+    this.failures = [...failures];
+    this.cause = failures.find((failure) => failure.error !== undefined)?.error;
+  }
+}
+
+export function isTelegramMessageDispatchReplayForgetError(
+  error: unknown,
+): error is TelegramMessageDispatchReplayForgetError {
+  return error instanceof TelegramMessageDispatchReplayForgetError;
+}
+
 function sanitizeFileSegment(value: string): string {
   const trimmed = value.trim();
   if (!trimmed) {
@@ -131,6 +158,30 @@ export async function commitTelegramMessageDispatchReplay(params: {
   );
 }
 
+export async function forgetTelegramMessageDispatchReplay(params: {
+  guard: TelegramMessageDispatchReplayGuard;
+  keys?: readonly string[];
+}): Promise<void> {
+  const keys = normalizeReplayKeys(params.keys);
+  const failures = (
+    await Promise.all(
+      keys.map(async (key): Promise<TelegramMessageDispatchReplayForgetFailure | null> => {
+        try {
+          const forgotten = await params.guard.forget(key, {
+            namespace: TELEGRAM_MESSAGE_DISPATCH_DEDUPE_NAMESPACE,
+          });
+          return forgotten ? null : { key };
+        } catch (error) {
+          return { key, error };
+        }
+      }),
+    )
+  ).filter((failure): failure is TelegramMessageDispatchReplayForgetFailure => Boolean(failure));
+  if (failures.length > 0) {
+    throw new TelegramMessageDispatchReplayForgetError(failures);
+  }
+}
+
 export function releaseTelegramMessageDispatchReplay(params: {
   guard: TelegramMessageDispatchReplayGuard;
   keys?: readonly string[];

extensions/telegram/src/network-errors.ts

@@ -225,7 +225,8 @@ function hasTelegramErrorCode(err: unknown, matches: (code: number) => boolean):
   return false;
 }
 
-function hasTelegramRetryAfter(err: unknown): boolean {
+/** Reads Telegram's flood-control retry_after hint (in ms) from any error nesting shape. */
+export function readTelegramRetryAfterMs(err: unknown): number | undefined {
   for (const candidate of collectTelegramErrorCandidates(err)) {
     if (!candidate || typeof candidate !== "object") {
       continue;
@@ -250,10 +251,10 @@ function hasTelegramRetryAfter(err: unknown): boolean {
                 ?.retry_after
             : undefined;
     if (typeof retryAfter === "number" && Number.isFinite(retryAfter)) {
-      return true;
+      return retryAfter * 1000;
     }
   }
-  return false;
+  return undefined;
 }
 
 /** Returns true for HTTP 5xx server errors (error may have been processed). */
@@ -264,10 +265,32 @@ export function isTelegramServerError(err: unknown): boolean {
 export function isTelegramRateLimitError(err: unknown): boolean {
   return (
     hasTelegramErrorCode(err, (code) => code === 429) ||
-    (hasTelegramRetryAfter(err) && /(?:^|\b)429\b|too many requests/i.test(formatErrorMessage(err)))
+    (readTelegramRetryAfterMs(err) !== undefined &&
+      /(?:^|\b)429\b|too many requests/i.test(formatErrorMessage(err)))
   );
 }
 
+const MESSAGE_NOT_MODIFIED_RE =
+  /400:\s*Bad Request:\s*message is not modified|MESSAGE_NOT_MODIFIED/i;
+const MESSAGE_HAS_NO_TEXT_RE = /400:\s*Bad Request:\s*there is no text in the message to edit/i;
+const EDIT_TARGET_MISSING_RE =
+  /400:\s*Bad Request:\s*message to edit not found|400:\s*Bad Request:\s*message can't be edited|MESSAGE_ID_INVALID/i;
+
+/** True when Telegram rejected an edit because the content is unchanged; the message already shows the requested text. */
+export function isTelegramMessageNotModifiedError(err: unknown): boolean {
+  return MESSAGE_NOT_MODIFIED_RE.test(formatErrorMessage(err));
+}
+
+/** True when the edit target has no text body (e.g. media message needing a caption edit). */
+export function isTelegramMessageHasNoTextError(err: unknown): boolean {
+  return MESSAGE_HAS_NO_TEXT_RE.test(formatErrorMessage(err));
+}
+
+/** True when the edit target is gone or locked (deleted message, invalid id); retrying the same edit cannot succeed. */
+export function isTelegramEditTargetMissingError(err: unknown): boolean {
+  return EDIT_TARGET_MISSING_RE.test(formatErrorMessage(err));
+}
+
 /** Returns true for HTTP 4xx client errors (Telegram explicitly rejected, not applied). */
 export function isTelegramClientRejection(err: unknown): boolean {
   return hasTelegramErrorCode(err, (code) => code >= 400 && code < 500);

extensions/telegram/src/polling-session.test.ts

@@ -73,6 +73,12 @@ let listTelegramSpooledUpdateClaims: typeof import("./telegram-ingress-spool.js"
 let listTelegramSpooledUpdates: typeof import("./telegram-ingress-spool.js").listTelegramSpooledUpdates;
 let recoverStaleTelegramSpooledUpdateClaims: typeof import("./telegram-ingress-spool.js").recoverStaleTelegramSpooledUpdateClaims;
 let writeTelegramSpooledUpdate: typeof import("./telegram-ingress-spool.js").writeTelegramSpooledUpdate;
+let createTelegramSpooledReplayDeferredParticipant: typeof import("./bot-processing-outcome.js").createTelegramSpooledReplayDeferredParticipant;
+let TelegramMessageDispatchReplayForgetError: typeof import("./message-dispatch-dedupe.js").TelegramMessageDispatchReplayForgetError;
+type TelegramMessageProcessingResult =
+  import("./bot-processing-outcome.js").TelegramMessageProcessingResult;
+type TelegramSpooledReplayDeferredParticipant =
+  import("./bot-processing-outcome.js").TelegramSpooledReplayDeferredParticipant;
 let beginTelegramReplyFence: typeof import("./telegram-reply-fence.js").beginTelegramReplyFence;
 let buildTelegramReplyFenceLaneKey: typeof import("./telegram-reply-fence.js").buildTelegramReplyFenceLaneKey;
 let endTelegramReplyFence: typeof import("./telegram-reply-fence.js").endTelegramReplyFence;
@@ -104,6 +110,7 @@ type WorkerPollSuccessListener = (message: {
 type WorkerPollErrorListener = (message: {
   type: "poll-error";
   message: string;
+  errorCode?: number;
   finishedAt: number;
 }) => void;
 type WorkerMessageListener = (message: TelegramIngressWorkerMessage) => void;
@@ -611,6 +618,9 @@ describe("TelegramPollingSession", () => {
       recoverStaleTelegramSpooledUpdateClaims,
       writeTelegramSpooledUpdate,
     } = await import("./telegram-ingress-spool.js"));
+    ({ createTelegramSpooledReplayDeferredParticipant } =
+      await import("./bot-processing-outcome.js"));
+    ({ TelegramMessageDispatchReplayForgetError } = await import("./message-dispatch-dedupe.js"));
     ({
       beginTelegramReplyFence,
       buildTelegramReplyFenceLaneKey,
@@ -1341,6 +1351,187 @@ describe("TelegramPollingSession", () => {
     });
   });
 
+  it("holds buffered spooled claims until deferred processing settles without blocking same-lane buffering", async () => {
+    await withTempSpool(async (tempDir) => {
+      const abort = new AbortController();
+      const participants: TelegramSpooledReplayDeferredParticipant[] = [];
+      const events: string[] = [];
+      await writeSpooledTestUpdates(tempDir, [
+        topicUpdate(42, 10, "first buffered topic 10 turn"),
+        topicUpdate(43, 10, "second buffered topic 10 turn"),
+      ]);
+
+      const { runPromise, stopWorker } = startIsolatedIngressSession({
+        abort,
+        spoolDir: tempDir,
+        drainIntervalMs: 10,
+        handleUpdate: async (update) => {
+          events.push(`topic10:${update.update_id}`);
+          const participant = createTelegramSpooledReplayDeferredParticipant(
+            `test-buffer:${update.update_id}`,
+          );
+          if (!participant) {
+            throw new Error("expected spooled replay participant");
+          }
+          participants.push(participant);
+        },
+      });
+
+      await vi.waitFor(() => expect(events).toEqual(["topic10:42", "topic10:43"]));
+      await vi.waitFor(async () =>
+        expect(
+          (await listTelegramSpooledUpdateClaims({ spoolDir: tempDir })).map(
+            (claim) => claim.updateId,
+          ),
+        ).toEqual([42, 43]),
+      );
+      expect(await pendingUpdateIds(tempDir, "all")).toEqual([]);
+
+      const completed: TelegramMessageProcessingResult = { kind: "completed" };
+      for (const participant of participants) {
+        participant.settle(completed);
+      }
+      await vi.waitFor(async () =>
+        expect(await listTelegramSpooledUpdateClaims({ spoolDir: tempDir })).toEqual([]),
+      );
+      expect(await pendingUpdateIds(tempDir, "all")).toEqual([]);
+      abort.abort();
+      stopWorker();
+      await runPromise;
+    });
+  });
+
+  it("releases buffered spooled claims for retry when deferred processing fails", async () => {
+    await withTempSpool(async (tempDir) => {
+      const abort = new AbortController();
+      const log = vi.fn();
+      const participants: TelegramSpooledReplayDeferredParticipant[] = [];
+      await writeSpooledTestUpdates(tempDir, [topicUpdate(42, 10, "buffered failure")]);
+
+      const { runPromise, stopWorker } = startIsolatedIngressSession({
+        abort,
+        spoolDir: tempDir,
+        log,
+        drainIntervalMs: 10,
+        handleUpdate: async (update) => {
+          const participant = createTelegramSpooledReplayDeferredParticipant(
+            `test-buffer:${update.update_id}`,
+          );
+          if (!participant) {
+            throw new Error("expected spooled replay participant");
+          }
+          participants.push(participant);
+        },
+      });
+
+      await vi.waitFor(() => expect(participants).toHaveLength(1));
+      await vi.waitFor(async () =>
+        expect(
+          (await listTelegramSpooledUpdateClaims({ spoolDir: tempDir })).map(
+            (claim) => claim.updateId,
+          ),
+        ).toEqual([42]),
+      );
+
+      abort.abort();
+      participants[0]?.settle({
+        kind: "failed-retryable",
+        error: new Error("buffered dispatch failed"),
+      });
+      await vi.waitFor(async () => expect(await pendingUpdateIds(tempDir, "all")).toEqual([42]));
+      expect(await listTelegramSpooledUpdateClaims({ spoolDir: tempDir })).toEqual([]);
+      expectLogIncludes(log, "spooled update 42 failed; keeping for retry");
+      stopWorker();
+      await runPromise;
+    });
+  });
+
+  it("dead-letters buffered spooled claims when dispatch dedupe rollback fails", async () => {
+    await withTempSpool(async (tempDir) => {
+      const abort = new AbortController();
+      const log = vi.fn();
+      const participants: TelegramSpooledReplayDeferredParticipant[] = [];
+      const events: string[] = [];
+      let attempts = 0;
+      await writeSpooledTestUpdates(tempDir, [topicUpdate(42, 10, "buffered rollback failure")]);
+
+      const { runPromise, stopWorker } = startIsolatedIngressSession({
+        abort,
+        spoolDir: tempDir,
+        log,
+        drainIntervalMs: 10,
+        handleUpdate: async (update) => {
+          attempts += 1;
+          if (attempts === 1) {
+            events.push(`dispatch:${update.update_id}`);
+            const participant = createTelegramSpooledReplayDeferredParticipant(
+              `test-buffer:${update.update_id}`,
+            );
+            if (!participant) {
+              throw new Error("expected spooled replay participant");
+            }
+            participants.push(participant);
+            return;
+          }
+          events.push(`duplicate-skip:${update.update_id}`);
+        },
+      });
+
+      await vi.waitFor(() => expect(participants).toHaveLength(1));
+      participants[0]?.settle({
+        kind: "failed-retryable",
+        error: new TelegramMessageDispatchReplayForgetError([{ key: "committed-dispatch-key" }]),
+      });
+
+      await vi.waitFor(async () => expect(await failedUpdateIds(tempDir)).toEqual([42]));
+      expect(events).toEqual(["dispatch:42"]);
+      expect(await pendingUpdateIds(tempDir, "all")).toEqual([]);
+      expect(await listTelegramSpooledUpdateClaims({ spoolDir: tempDir })).toEqual([]);
+      expectLogIncludes(log, "non-retryable dispatch-dedupe-rollback-failed");
+      expectLogExcludes(log, "spooled update 42 failed; keeping for retry");
+
+      abort.abort();
+      stopWorker();
+      await runPromise;
+    });
+  });
+
+  it("fails buffered spooled claims instead of requeueing when deferred processing times out", async () => {
+    await withTempSpool(async (tempDir) => {
+      const abort = new AbortController();
+      const log = vi.fn();
+      const participants: TelegramSpooledReplayDeferredParticipant[] = [];
+      await writeSpooledTestUpdates(tempDir, [topicUpdate(42, 10, "buffered timeout")]);
+
+      const { runPromise, stopWorker } = startIsolatedIngressSession({
+        abort,
+        spoolDir: tempDir,
+        log,
+        drainIntervalMs: 10,
+        spooledUpdateHandlerTimeoutMs: 20,
+        handleUpdate: async (update) => {
+          const participant = createTelegramSpooledReplayDeferredParticipant(
+            `test-buffer:${update.update_id}`,
+          );
+          if (!participant) {
+            throw new Error("expected spooled replay participant");
+          }
+          participants.push(participant);
+        },
+      });
+
+      await vi.waitFor(() => expect(participants).toHaveLength(1));
+      await vi.waitFor(async () => expect(await failedUpdateIds(tempDir)).toEqual([42]));
+      expect(await pendingUpdateIds(tempDir, "all")).toEqual([]);
+      expect(await listTelegramSpooledUpdateClaims({ spoolDir: tempDir })).toEqual([]);
+      expectLogIncludes(log, "buffered processing timed out behind update 42");
+      expectLogExcludes(log, "spooled update 42 failed; keeping for retry");
+      abort.abort();
+      stopWorker();
+      await runPromise;
+    });
+  });
+
   it("dead-letters missing harness failures so later same-lane updates can drain", async () => {
     await withTempSpool(async (tempDir) => {
       const abort = new AbortController();
@@ -2356,6 +2547,89 @@ describe("TelegramPollingSession", () => {
     }
   });
 
+  it("restarts isolated ingress on a getUpdates conflict instead of crashing the account", async () => {
+    const abort = new AbortController();
+    const tempDir = await fs.mkdtemp(path.join(os.tmpdir(), "openclaw-telegram-spool-"));
+    const log = vi.fn();
+    const setStatus = vi.fn();
+    // 409 conflicts are not "recoverable network errors"; the conflict branch
+    // must restart the cycle before that classifier is consulted.
+    isRecoverableTelegramNetworkErrorMock.mockReturnValue(false);
+    const deleteWebhook = vi.fn(async () => true);
+    createTelegramBotMock.mockImplementation(() => ({
+      api: {
+        deleteWebhook,
+        config: { use: vi.fn() },
+      },
+      init: vi.fn(async () => undefined),
+      handleUpdate: vi.fn(async () => undefined),
+      stop: vi.fn(async () => undefined),
+    }));
+    const transport1 = makeTelegramTransport();
+    const transport2 = makeTelegramTransport();
+    const createTelegramTransport = vi
+      .fn<() => ReturnType<typeof makeTelegramTransport>>()
+      .mockReturnValueOnce(transport2);
+
+    let workerCycle = 0;
+    let listener: WorkerPollErrorListener | undefined;
+    const createWorker = vi.fn(() => ({
+      onMessage: vi.fn((next: WorkerPollErrorListener) => {
+        listener = next;
+        return () => undefined;
+      }),
+      stop: vi.fn(async () => undefined),
+      task: vi.fn(async () => {
+        workerCycle += 1;
+        if (workerCycle === 1) {
+          listener?.({
+            type: "poll-error",
+            message: "Conflict: terminated by other getUpdates request",
+            errorCode: 409,
+            finishedAt: Date.now(),
+          });
+          throw new Error("Telegram ingress worker exited with code 1");
+        }
+        abort.abort();
+      }),
+    }));
+
+    try {
+      const session = createPollingSession({
+        abortSignal: abort.signal,
+        log,
+        setStatus,
+        telegramTransport: transport1,
+        createTelegramTransport,
+        isolatedIngress: {
+          enabled: true,
+          spoolDir: tempDir,
+          createWorker,
+          drainIntervalMs: 100,
+        },
+      });
+
+      await session.runUntilAbort();
+
+      expect(createWorker).toHaveBeenCalledTimes(2);
+      // The conflict resets webhook cleanup so the next cycle re-runs deleteWebhook.
+      expect(deleteWebhook).toHaveBeenCalledTimes(2);
+      // The conflict marks the transport dirty so the next cycle gets a fresh socket.
+      expect(createTelegramTransport).toHaveBeenCalledTimes(1);
+      expectLogIncludes(log, "Another OpenClaw gateway, script, or Telegram poller");
+      expect(
+        statusPatches(setStatus).some(
+          (patch) =>
+            patch.connected === false &&
+            String(patch.lastError).includes("Another OpenClaw gateway"),
+        ),
+      ).toBe(true);
+    } finally {
+      abort.abort();
+      await fs.rm(tempDir, { recursive: true, force: true });
+    }
+  });
+
   it("keeps active spooled lanes blocked across account restarts", async () => {
     vi.useFakeTimers({ shouldAdvanceTime: true });
     const firstAbort = new AbortController();
@@ -3533,9 +3807,11 @@ describe("TelegramPollingSession", () => {
     const watchdogHarness = installPollingStallWatchdogHarness();
 
     const log = vi.fn();
+    const setStatus = vi.fn();
     const session = createPollingSession({
       abortSignal: abort.signal,
       log,
+      setStatus,
     });
 
     try {
@@ -3567,6 +3843,14 @@ describe("TelegramPollingSession", () => {
       abort.abort();
       resolveFirstTask();
       await runPromise;
+
+      // The stall must reach channel status, not just the gateway log.
+      expect(
+        statusPatches(setStatus).some(
+          (patch) =>
+            patch.connected === false && String(patch.lastError).includes("Polling stall detected"),
+        ),
+      ).toBe(true);
     } finally {
       watchdogHarness.restore();
     }
@@ -3614,6 +3898,7 @@ describe("TelegramPollingSession", () => {
   it("logs an actionable duplicate-poller hint for getUpdates conflicts", async () => {
     const abort = new AbortController();
     const log = vi.fn();
+    const setStatus = vi.fn();
     const conflictError = Object.assign(
       new Error("Conflict: terminated by other getUpdates request"),
       {
@@ -3628,11 +3913,19 @@ describe("TelegramPollingSession", () => {
     const session = createPollingSession({
       abortSignal: abort.signal,
       log,
+      setStatus,
     });
 
     await session.runUntilAbort();
 
     expectLogIncludes(log, "Another OpenClaw gateway, script, or Telegram poller");
+    // The hint must reach channel status, not just the gateway log.
+    expect(
+      statusPatches(setStatus).some(
+        (patch) =>
+          patch.connected === false && String(patch.lastError).includes("Another OpenClaw gateway"),
+      ),
+    ).toBe(true);
   });
 
   it("logs polling cycle start after a transport rebuild", async () => {

extensions/telegram/src/polling-session.ts

@@ -19,8 +19,14 @@ import {
 } from "openclaw/plugin-sdk/runtime-env";
 import { normalizeLowercaseStringOrEmpty } from "openclaw/plugin-sdk/string-coerce-runtime";
 import { withTelegramApiErrorLogging } from "./api-logging.js";
+import {
+  runWithTelegramSpooledReplayUpdate,
+  type TelegramMessageProcessingResult,
+  type TelegramSpooledReplayDeferredParticipant,
+} from "./bot-processing-outcome.js";
 import { createTelegramBot } from "./bot.js";
 import type { TelegramTransport } from "./fetch.js";
+import { isTelegramMessageDispatchReplayForgetError } from "./message-dispatch-dedupe.js";
 import { isRecoverableTelegramNetworkError } from "./network-errors.js";
 import { TelegramPollingLivenessTracker } from "./polling-liveness.js";
 import { createTelegramPollingStatusPublisher } from "./polling-status.js";
@@ -110,6 +116,11 @@ function resolveTelegramRestartDelayMs(
   return { delayMs, stopTimeoutSuffix };
 }
 
+// Surfaced in logs and channel status when getUpdates returns 409; the only
+// user-fixable causes are a second poller on the same token or a stale webhook.
+const TELEGRAM_GET_UPDATES_CONFLICT_HINT =
+  " Another OpenClaw gateway, script, or Telegram poller may be using this bot token; stop the duplicate poller or switch this account to webhook mode.";
+
 const DEFAULT_POLL_STALL_THRESHOLD_MS = 120_000;
 const MIN_POLL_STALL_THRESHOLD_MS = 30_000;
 const MAX_POLL_STALL_THRESHOLD_MS = 600_000;
@@ -131,7 +142,7 @@ function normalizeTelegramAccountId(accountId?: string | null): string {
 }
 
 type NonRetryableSpooledUpdateFailure = {
-  reason: "missing-agent-harness";
+  reason: "missing-agent-harness" | "dispatch-dedupe-rollback-failed";
   message: string;
 };
 
@@ -143,6 +154,11 @@ function resolveNonRetryableSpooledUpdateFailure(
     current.error,
   ])) {
     const message = formatErrorMessage(candidate);
+    if (isTelegramMessageDispatchReplayForgetError(candidate)) {
+      // A committed dispatch key that cannot be rolled back makes retry unsafe:
+      // the next replay can be duplicate-suppressed and then deleted.
+      return { reason: "dispatch-dedupe-rollback-failed", message };
+    }
     if (
       readErrorName(candidate) === MISSING_AGENT_HARNESS_ERROR_NAME ||
       MISSING_AGENT_HARNESS_MESSAGE_RE.test(message)
@@ -258,6 +274,22 @@ type SpooledUpdateHandlerState = {
   timeoutMessage?: string;
 };
 
+type DeferredSpooledUpdateClaimState = {
+  claimKey: string;
+  laneKey: string;
+  task: Promise<void>;
+  timer?: ReturnType<typeof setTimeout>;
+  timedOutMessage?: string;
+  update: ClaimedTelegramSpooledUpdate;
+  updateId: number;
+};
+
+const deferredSpooledUpdateClaimsByKey = new Map<string, DeferredSpooledUpdateClaimState>();
+
+function buildDeferredSpooledUpdateClaimKey(update: ClaimedTelegramSpooledUpdate): string {
+  return `${update.pendingPath}:${update.claim?.claimToken ?? update.claim?.processId ?? "claimed"}`;
+}
+
 type SpooledUpdateDrainResult = {
   blockedByLane: Set<string>;
   started: number;
@@ -299,6 +331,7 @@ export class TelegramPollingSession {
   #activeRunner: ReturnType<typeof run> | undefined;
   #activeFetchAbort: AbortController | undefined;
   #spooledUpdateHandlerKeys = new Set<string>();
+  #deferredSpooledUpdateClaimKeys = new Set<string>();
   #transportState: TelegramPollingTransportState;
   #status: ReturnType<typeof createTelegramPollingStatusPublisher>;
   #stallThresholdMs: number;
@@ -522,10 +555,12 @@ export class TelegramPollingSession {
     bot: TelegramBot;
     update: ClaimedTelegramSpooledUpdate;
   }): Promise<boolean> {
+    let replay: { deferredWork?: TelegramSpooledReplayDeferredParticipant };
     try {
-      await params.bot.handleUpdate(
-        params.update.update as Parameters<typeof params.bot.handleUpdate>[0],
-      );
+      const update = params.update.update as Parameters<typeof params.bot.handleUpdate>[0];
+      replay = await runWithTelegramSpooledReplayUpdate(update, async () => {
+        await params.bot.handleUpdate(update);
+      });
     } catch (err) {
       await this.#releaseFailedSpooledUpdate({
         err,
@@ -533,6 +568,14 @@ export class TelegramPollingSession {
       });
       return false;
     }
+    if (replay.deferredWork) {
+      this.#registerDeferredSpooledUpdate({
+        deferredWork: replay.deferredWork,
+        laneKey: this.#spooledUpdateLaneKey(params.update),
+        update: params.update,
+      });
+      return true;
+    }
     try {
       await deleteTelegramSpooledUpdate(params.update);
       return true;
@@ -544,6 +587,115 @@ export class TelegramPollingSession {
     }
   }
 
+  #registerDeferredSpooledUpdate(params: {
+    deferredWork: TelegramSpooledReplayDeferredParticipant;
+    laneKey: string;
+    update: ClaimedTelegramSpooledUpdate;
+  }): void {
+    const claimKey = buildDeferredSpooledUpdateClaimKey(params.update);
+    const previous = deferredSpooledUpdateClaimsByKey.get(claimKey);
+    if (previous) {
+      if (previous.timer) {
+        clearTimeout(previous.timer);
+      }
+      deferredSpooledUpdateClaimsByKey.delete(claimKey);
+    }
+    let settled = false;
+    const finish = async (result: TelegramMessageProcessingResult): Promise<void> => {
+      if (settled) {
+        return;
+      }
+      settled = true;
+      if (state.timer) {
+        clearTimeout(state.timer);
+      }
+      if (deferredSpooledUpdateClaimsByKey.get(claimKey) === state) {
+        deferredSpooledUpdateClaimsByKey.delete(claimKey);
+      }
+      this.#deferredSpooledUpdateClaimKeys.delete(claimKey);
+      if (result.kind === "failed-retryable") {
+        if (state.timedOutMessage) {
+          await this.#failTimedOutDeferredSpooledUpdate(state);
+          return;
+        }
+        await this.#releaseFailedSpooledUpdate({
+          err: result.error,
+          update: params.update,
+        });
+        return;
+      }
+      try {
+        await deleteTelegramSpooledUpdate(params.update);
+      } catch (err) {
+        this.opts.log(
+          `[telegram][diag] spooled update ${params.update.updateId} completed after buffered processing but processing marker cleanup failed: ${formatErrorMessage(err)}`,
+        );
+      }
+    };
+    const state: DeferredSpooledUpdateClaimState = {
+      claimKey,
+      laneKey: params.laneKey,
+      task: params.deferredWork.task.then(finish, async (err: unknown) => {
+        await finish({ kind: "failed-retryable", error: err });
+      }),
+      update: params.update,
+      updateId: params.update.updateId,
+    };
+    state.timer = setTimeout(() => {
+      const age = formatDurationPrecise(this.#spooledUpdateHandlerTimeoutMs);
+      state.timedOutMessage = `Telegram isolated polling spool buffered processing timed out behind update ${params.update.updateId} on lane ${params.laneKey} after ${age}; marking the update failed, aborting active reply work, and keeping the claim out of retry while the buffered task settles.`;
+      params.deferredWork.settle({
+        kind: "failed-retryable",
+        error: new Error(state.timedOutMessage),
+      });
+    }, this.#spooledUpdateHandlerTimeoutMs);
+    state.timer.unref?.();
+    deferredSpooledUpdateClaimsByKey.set(claimKey, state);
+    this.#deferredSpooledUpdateClaimKeys.add(claimKey);
+  }
+
+  #isDeferredSpooledUpdateClaim(update: ClaimedTelegramSpooledUpdate): boolean {
+    return deferredSpooledUpdateClaimsByKey.has(buildDeferredSpooledUpdateClaimKey(update));
+  }
+
+  async #failTimedOutDeferredSpooledUpdate(state: DeferredSpooledUpdateClaimState): Promise<void> {
+    const message =
+      state.timedOutMessage ??
+      `Telegram isolated polling spool buffered processing timed out behind update ${state.updateId} on lane ${state.laneKey}; marking the update failed.`;
+    try {
+      const failed = await failTelegramSpooledUpdateClaim({
+        update: state.update,
+        reason: "handler-timeout",
+        message,
+      });
+      if (!failed) {
+        this.opts.log(
+          `[telegram][diag] timed out buffered spooled update ${state.updateId} no longer had a processing marker to fail.`,
+        );
+        this.#status.notePollingError(message);
+        return;
+      }
+    } catch (err) {
+      this.opts.log(
+        `[telegram][diag] timed out buffered spooled update ${state.updateId} could not be marked failed: ${formatErrorMessage(err)}`,
+      );
+      this.#status.notePollingError(message);
+      return;
+    }
+    const scopedReplyFenceLaneKey = buildTelegramReplyFenceLaneKey({
+      accountId: this.opts.accountId,
+      sequentialKey: state.laneKey,
+    });
+    const abortedReplyWork = supersedeTelegramReplyFenceLane(scopedReplyFenceLaneKey);
+    if (!abortedReplyWork) {
+      this.opts.log(
+        `[telegram][diag] timed out buffered spooled update ${state.updateId} had no active reply fence on lane ${state.laneKey}.`,
+      );
+    }
+    this.opts.log(`[telegram] ${message}`);
+    this.#status.notePollingError(message);
+  }
+
   async #releaseFailedSpooledUpdate(params: {
     err: unknown;
     update: ClaimedTelegramSpooledUpdate;
@@ -586,11 +738,14 @@ export class TelegramPollingSession {
   }
 
   async #waitForSpooledUpdateHandlers(): Promise<void> {
-    await Promise.allSettled(
-      [...this.#spooledUpdateHandlerKeys]
+    await Promise.allSettled([
+      ...[...this.#spooledUpdateHandlerKeys]
         .map((handlerKey) => activeSpooledUpdateHandlersByLane.get(handlerKey)?.task)
         .filter((task): task is Promise<boolean> => Boolean(task)),
-    );
+      ...[...this.#deferredSpooledUpdateClaimKeys]
+        .map((claimKey) => deferredSpooledUpdateClaimsByKey.get(claimKey)?.task)
+        .filter((task): task is Promise<void> => Boolean(task)),
+    ]);
   }
 
   #spooledUpdateLaneKey(update: TelegramSpooledUpdate): string {
@@ -619,6 +774,7 @@ export class TelegramPollingSession {
       spoolDir: params.spoolDir,
       staleMs: 0,
       shouldRecover: (claim) =>
+        !this.#isDeferredSpooledUpdateClaim(claim) &&
         !activeLaneKeys.has(this.#spooledUpdateLaneKey(claim)) &&
         !isTelegramSpooledUpdateClaimOwnedByOtherLiveProcess(claim),
     });
@@ -627,7 +783,9 @@ export class TelegramPollingSession {
         await listTelegramSpooledUpdateClaims({
           spoolDir: params.spoolDir,
         })
-      ).map((claim) => this.#spooledUpdateLaneKey(claim)),
+      )
+        .filter((claim) => !this.#isDeferredSpooledUpdateClaim(claim))
+        .map((claim) => this.#spooledUpdateLaneKey(claim)),
     );
     const updates = await listTelegramSpooledUpdates({
       spoolDir: params.spoolDir,
@@ -813,10 +971,12 @@ export class TelegramPollingSession {
       offset: number | null;
       outcome: string;
       error?: string;
+      errorCode: number | null;
     } = {
       startedAt: null,
       offset: null,
       outcome: "not-started",
+      errorCode: null,
     };
     const liveness = new TelegramPollingLivenessTracker();
     let consecutiveDrainFailures = 0;
@@ -853,6 +1013,7 @@ export class TelegramPollingSession {
         pollState.offset = message.offset;
         pollState.outcome = "started";
         delete pollState.error;
+        pollState.errorCode = null;
         return;
       }
       if (message.type === "poll-success") {
@@ -871,6 +1032,7 @@ export class TelegramPollingSession {
         liveness.noteGetUpdatesFinished();
         pollState.outcome = "error";
         pollState.error = message.message;
+        pollState.errorCode = message.errorCode ?? null;
         return;
       }
       if (message.type === "update") {
@@ -1002,14 +1164,24 @@ export class TelegramPollingSession {
         if (this.opts.abortSignal?.aborted) {
           return "exit";
         }
-        if (
+        // The worker only issues getUpdates, so a 409 is always a duplicate
+        // poller (or stale webhook) conflict. Mirror the classic polling
+        // cycle: re-clear the webhook, rotate the transport (#69787), and
+        // restart with backoff instead of crashing the whole account.
+        const isConflict = pollState.errorCode === 409;
+        if (isConflict) {
+          this.#webhookCleared = false;
+          this.#transportState.markDirty();
+        } else if (
           pollState.error &&
           !isRecoverableTelegramNetworkError(new Error(pollState.error), { context: "polling" })
         ) {
           this.#status.notePollingError(pollState.error);
           throw new Error(pollState.error, { cause: err });
         }
-        const message = formatErrorMessage(err);
+        const message = isConflict
+          ? `Telegram getUpdates conflict: ${pollState.error}.${TELEGRAM_GET_UPDATES_CONFLICT_HINT}`
+          : formatErrorMessage(err);
         this.opts.log(`[telegram][diag] isolated polling ingress failed: ${message}`);
         this.#status.notePollingError(message);
         clearForceCycleTimer();
@@ -1162,6 +1334,7 @@ export class TelegramPollingSession {
         this.#transportState.markDirty();
         stalledRestart = true;
         this.opts.log(`[telegram] ${stall.message}`);
+        this.#status.notePollingError(stall.message);
         requestStopForRestart();
       }
     }, POLL_WATCHDOG_INTERVAL_MS);
@@ -1210,12 +1383,16 @@ export class TelegramPollingSession {
       }
       const reason = isConflict ? "getUpdates conflict" : "network error";
       const errMsg = formatErrorMessage(err);
-      const conflictHint = isConflict
-        ? " Another OpenClaw gateway, script, or Telegram poller may be using this bot token; stop the duplicate poller or switch this account to webhook mode."
-        : "";
+      const conflictHint = isConflict ? TELEGRAM_GET_UPDATES_CONFLICT_HINT : "";
       this.opts.log(
         `[telegram][diag] polling cycle error reason=${reason} ${liveness.formatDiagnosticFields("lastGetUpdatesError")} err=${errMsg}${conflictHint}`,
       );
+      // Conflicts carry a user-fixable diagnosis, so surface them in channel
+      // status. Recoverable network blips stay log-only; the stall watchdog
+      // owns status for extended outages (see detectStall above).
+      if (isConflict) {
+        this.#status.notePollingError(`Telegram ${reason}: ${errMsg}.${conflictHint}`);
+      }
       clearForceCycleTimer();
       const shouldRestart = await this.#waitBeforeRestart(
         (delay) => `Telegram ${reason}: ${errMsg};${conflictHint} retrying in ${delay}.`,

extensions/telegram/src/send.ts

@@ -29,6 +29,8 @@ import { buildInlineKeyboard } from "./inline-keyboard.js";
 import {
   isRecoverableTelegramNetworkError,
   isSafeToRetrySendError,
+  isTelegramMessageHasNoTextError,
+  isTelegramMessageNotModifiedError,
   isTelegramRateLimitError,
   isTelegramServerError,
 } from "./network-errors.js";
@@ -230,9 +232,6 @@ function logTelegramOutboundSendOk(params: TelegramOutboundSuccessLogParams): vo
 }
 
 const PARSE_ERR_RE = /can't parse entities|parse entities|find end of the entity/i;
-const MESSAGE_NOT_MODIFIED_RE =
-  /400:\s*Bad Request:\s*message is not modified|MESSAGE_NOT_MODIFIED/i;
-const MESSAGE_HAS_NO_TEXT_RE = /400:\s*Bad Request:\s*there is no text in the message to edit/i;
 const MESSAGE_DELETE_NOOP_RE =
   /message to delete not found|message can't be deleted|MESSAGE_ID_INVALID|MESSAGE_DELETE_FORBIDDEN/i;
 const CHAT_NOT_FOUND_RE = /400: Bad Request: chat not found/i;
@@ -422,14 +421,6 @@ function normalizeMessageId(raw: string | number): number {
   throw new Error("Message id is required for Telegram actions");
 }
 
-function isTelegramMessageNotModifiedError(err: unknown): boolean {
-  return MESSAGE_NOT_MODIFIED_RE.test(formatErrorMessage(err));
-}
-
-function isTelegramMessageHasNoTextError(err: unknown): boolean {
-  return MESSAGE_HAS_NO_TEXT_RE.test(formatErrorMessage(err));
-}
-
 function isTelegramMessageDeleteNoopError(err: unknown): boolean {
   return MESSAGE_DELETE_NOOP_RE.test(formatErrorMessage(err));
 }

extensions/telegram/src/telegram-ingress-worker.runtime.ts

@@ -51,6 +51,26 @@ function formatErrorMessage(err: unknown): string {
   return String(err);
 }
 
+function readTelegramErrorCode(err: unknown): number | undefined {
+  if (err && typeof err === "object" && "error_code" in err) {
+    const code = (err as { error_code: unknown }).error_code;
+    if (typeof code === "number") {
+      return code;
+    }
+  }
+  return undefined;
+}
+
+function postPollError(err: unknown): void {
+  const errorCode = readTelegramErrorCode(err);
+  post({
+    type: "poll-error",
+    message: formatErrorMessage(err),
+    ...(errorCode === undefined ? {} : { errorCode }),
+    finishedAt: Date.now(),
+  });
+}
+
 function resolveBackoff(attempt: number): number {
   return Math.min(retryMaxMs, retryInitialMs * 2 ** Math.max(0, attempt - 1));
 }
@@ -122,15 +142,21 @@ async function fetchJson(params: {
     });
     const json = (await response.json()) as {
       ok?: unknown;
+      error_code?: unknown;
       result?: unknown;
       description?: unknown;
     };
     if (!response.ok || json.ok !== true) {
-      throw new Error(
+      const message =
         typeof json.description === "string"
           ? json.description
-          : `Telegram getUpdates failed with HTTP ${response.status}`,
-      );
+          : `Telegram getUpdates failed with HTTP ${response.status}`;
+      // Preserve the Bot API error_code across the worker boundary so the
+      // parent session can distinguish getUpdates conflicts (409) from fatal
+      // errors (401) without parsing description strings.
+      throw typeof json.error_code === "number"
+        ? Object.assign(new Error(message), { error_code: json.error_code })
+        : new Error(message);
     }
     return json.result;
   } finally {
@@ -195,11 +221,7 @@ async function main(): Promise<void> {
           break;
         }
         failures += 1;
-        post({
-          type: "poll-error",
-          message: formatErrorMessage(err),
-          finishedAt: Date.now(),
-        });
+        postPollError(err);
         if (!isRecoverableTelegramNetworkError(err, { context: "polling" })) {
           throw err;
         }
@@ -216,7 +238,7 @@ main()
     parentPort?.close();
   })
   .catch((err: unknown) => {
-    post({ type: "poll-error", message: formatErrorMessage(err), finishedAt: Date.now() });
+    postPollError(err);
     parentPort?.close();
     process.exitCode = stopped ? 0 : 1;
   });

extensions/telegram/src/telegram-ingress-worker.ts

@@ -17,6 +17,8 @@ export type TelegramIngressWorkerMessage =
   | {
       type: "poll-error";
       message: string;
+      /** Telegram Bot API error_code (e.g. 409 for getUpdates conflicts). */
+      errorCode?: number;
       finishedAt: number;
     }
   | {

package.json

@@ -1947,7 +1947,7 @@
     "@types/ws": "8.18.1",
     "@typescript/native-preview": "7.0.0-dev.20260527.2",
     "@vitest/coverage-v8": "4.1.7",
-    "esbuild": "0.28.0",
+    "esbuild": "0.28.1",
     "jscpd": "4.2.4",
     "jsdom": "29.1.1",
     "lit": "3.3.3",

packages/agent-core/src/agent-loop.test.ts

@@ -1,9 +1,17 @@
 // Agent Core tests cover agent loop behavior.
+import { Type } from "typebox";
 import { describe, expect, it } from "vitest";
 import { agentLoop, agentLoopContinue } from "./agent-loop.js";
 import { createAssistantMessageEventStream } from "./llm.js";
 import type { AssistantMessage, Message, Model } from "./llm.js";
-import type { AgentContext, AgentEvent, AgentLoopConfig, AgentMessage, StreamFn } from "./types.js";
+import type {
+  AgentContext,
+  AgentEvent,
+  AgentLoopConfig,
+  AgentMessage,
+  AgentTool,
+  StreamFn,
+} from "./types.js";
 
 const model: Model = {
   id: "test-model",
@@ -143,6 +151,146 @@ describe("agentLoop streaming updates", () => {
   });
 });
 
+describe("agentLoop tool termination", () => {
+  function makeAssistantMessage(content: AssistantMessage["content"]): AssistantMessage {
+    return {
+      role: "assistant",
+      content,
+      api: model.api,
+      provider: model.provider,
+      model: model.id,
+      usage: {
+        input: 0,
+        output: 0,
+        cacheRead: 0,
+        cacheWrite: 0,
+        totalTokens: 0,
+        cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, total: 0 },
+      },
+      stopReason: content.some((item) => item.type === "toolCall") ? "toolUse" : "stop",
+      timestamp: 1,
+    };
+  }
+
+  function makeTool(name: string, executed: string[]): AgentTool {
+    return {
+      name,
+      label: name,
+      description: name,
+      parameters: Type.Object({}, { additionalProperties: false }),
+      execute: async () => {
+        executed.push(name);
+        return {
+          content: [{ type: "text", text: `${name} result` }],
+          details: { name },
+        };
+      },
+    };
+  }
+
+  it("continues after a side-effect tool result when afterToolCall records it without terminate", async () => {
+    const executed: string[] = [];
+    let turn = 0;
+    const streamFn: StreamFn = () => {
+      turn += 1;
+      const stream = createAssistantMessageEventStream();
+      queueMicrotask(() => {
+        const message =
+          turn === 1
+            ? makeAssistantMessage([
+                { type: "toolCall", id: "call-message", name: "message", arguments: {} },
+              ])
+            : turn === 2
+              ? makeAssistantMessage([
+                  { type: "toolCall", id: "call-exec", name: "exec", arguments: {} },
+                ])
+              : makeAssistantMessage([{ type: "text", text: "done" }]);
+        stream.push({
+          type: "done",
+          reason: message.stopReason === "toolUse" ? "toolUse" : "stop",
+          message,
+        });
+        stream.end();
+      });
+      return stream;
+    };
+    let recordedSideEffect = false;
+
+    const stream = agentLoop(
+      [{ role: "user", content: "hello", timestamp: 1 }],
+      {
+        systemPrompt: "",
+        messages: [],
+        tools: [makeTool("message", executed), makeTool("exec", executed)],
+      },
+      {
+        ...config,
+        afterToolCall: async ({ toolCall }) => {
+          if (toolCall.name === "message") {
+            recordedSideEffect = true;
+          }
+          return undefined;
+        },
+      },
+      undefined,
+      streamFn,
+    );
+
+    const events = await collectEvents(stream);
+
+    expect(recordedSideEffect).toBe(true);
+    expect(turn).toBe(3);
+    expect(executed).toEqual(["message", "exec"]);
+    expect(events.filter((event) => event.type === "tool_execution_start")).toHaveLength(2);
+    expect(events.at(-1)).toMatchObject({ type: "agent_end" });
+  });
+
+  it("stops after a tool result only when the finalized result explicitly terminates", async () => {
+    const executed: string[] = [];
+    let turn = 0;
+    const streamFn: StreamFn = () => {
+      turn += 1;
+      const stream = createAssistantMessageEventStream();
+      queueMicrotask(() => {
+        const message =
+          turn === 1
+            ? makeAssistantMessage([
+                { type: "toolCall", id: "call-message", name: "message", arguments: {} },
+              ])
+            : makeAssistantMessage([
+                { type: "toolCall", id: "call-exec", name: "exec", arguments: {} },
+              ]);
+        stream.push({ type: "done", reason: "toolUse", message });
+        stream.end();
+      });
+      return stream;
+    };
+
+    const stream = agentLoop(
+      [{ role: "user", content: "hello", timestamp: 1 }],
+      {
+        systemPrompt: "",
+        messages: [],
+        tools: [makeTool("message", executed), makeTool("exec", executed)],
+      },
+      {
+        ...config,
+        afterToolCall: async ({ toolCall }) =>
+          toolCall.name === "message" ? { terminate: true } : undefined,
+      },
+      undefined,
+      streamFn,
+    );
+
+    const events = await collectEvents(stream);
+
+    expect(turn).toBe(1);
+    expect(executed).toEqual(["message"]);
+    expect(events.filter((event) => event.type === "tool_execution_start")).toHaveLength(1);
+    expect(events.at(-1)).toMatchObject({ type: "agent_end" });
+  });
+});
+
 describe("agentLoop thinking state", () => {
   function makeAssistantMessage(
     activeModel: Model,

packages/llm-core/src/types.ts

@@ -296,6 +296,9 @@ export interface AssistantMessage {
   usage: Usage;
   stopReason: StopReason;
   errorMessage?: string;
+  errorCode?: string;
+  errorType?: string;
+  errorBody?: string;
   timestamp: number; // Unix timestamp in milliseconds
 }
 

pnpm-lock.yaml

@@ -263,8 +263,8 @@ importers:
         specifier: 4.1.7
         version: 4.1.7(@vitest/browser@4.1.7)(vitest@4.1.7)
       esbuild:
-        specifier: 0.28.0
-        version: 0.28.0
+        specifier: 0.28.1
+        version: 0.28.1
       jscpd:
         specifier: 4.2.4
         version: 4.2.4
@@ -300,7 +300,7 @@ importers:
         version: 0.3.0
       vitest:
         specifier: 4.1.7
-        version: 4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.1)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))
+        version: 4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.1)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))
     optionalDependencies:
       sqlite-vec:
         specifier: 0.1.9
@@ -1942,7 +1942,7 @@ importers:
         version: 14.1.2
       '@vitest/browser-playwright':
         specifier: 4.1.7
-        version: 4.1.7(playwright@1.60.0)(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))(vitest@4.1.7)
+        version: 4.1.7(playwright@1.60.0)(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))(vitest@4.1.7)
       jsdom:
         specifier: 29.1.1
         version: 29.1.1(@noble/hashes@2.0.1)
@@ -1951,10 +1951,10 @@ importers:
         version: 1.60.0
       vite:
         specifier: 8.0.14
-        version: 8.0.14(@types/node@25.9.2)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0)
+        version: 8.0.14(@types/node@25.9.2)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0)
       vitest:
         specifier: 4.1.7
-        version: 4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.2)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))
+        version: 4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.2)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))
 
 packages:
 
@@ -2482,158 +2482,158 @@ packages:
   '@emnapi/wasi-threads@1.2.1':
     resolution: {integrity: sha512-uTII7OYF+/Mes/MrcIOYp5yOtSMLBWSIoLPpcgwipoiKbli6k322tcoFsxoIIxPDqW01SQGAgko4EzZi2BNv2w==}
 
-  '@esbuild/aix-ppc64@0.28.0':
-    resolution: {integrity: sha512-lhRUCeuOyJQURhTxl4WkpFTjIsbDayJHih5kZC1giwE+MhIzAb7mEsQMqMf18rHLsrb5qI1tafG20mLxEWcWlA==}
+  '@esbuild/aix-ppc64@0.28.1':
+    resolution: {integrity: sha512-Svl7tq8k/08+p6CXPpRjQ1fKX+1odH/BQbb48fV6fj3CWHhsoIOoY87w1oHXm0qEpkIK3ZfVgp0hed3XBXzXMQ==}
     engines: {node: '>=18'}
     cpu: [ppc64]
     os: [aix]
 
-  '@esbuild/android-arm64@0.28.0':
-    resolution: {integrity: sha512-+WzIXQOSaGs33tLEgYPYe/yQHf0WTU0X42Jca3y8NWMbUVhp7rUnw+vAsRC/QiDrdD31IszMrZy+qwPOPjd+rw==}
+  '@esbuild/android-arm64@0.28.1':
+    resolution: {integrity: sha512-34EGEbCIAgosYz6goLcopX6Mo7NyGv9tfwEM2/7Ce2VcVRk568iSvniGWcUXIy7wEDR1wzolcxcriFVrWYcwBg==}
     engines: {node: '>=18'}
     cpu: [arm64]
     os: [android]
 
-  '@esbuild/android-arm@0.28.0':
-    resolution: {integrity: sha512-wqh0ByljabXLKHeWXYLqoJ5jKC4XBaw6Hk08OfMrCRd2nP2ZQ5eleDZC41XHyCNgktBGYMbqnrJKq/K/lzPMSQ==}
+  '@esbuild/android-arm@0.28.1':
+    resolution: {integrity: sha512-0k2F129Xdio1TdJfzJ8sy1Q47vUD2NnwdhiAf7drUN1EBTfPf4hsFCtmMgu/6m8JSzsBrlmVjudMBQqOfG8usQ==}
     engines: {node: '>=18'}
     cpu: [arm]
     os: [android]
 
-  '@esbuild/android-x64@0.28.0':
-    resolution: {integrity: sha512-+VJggoaKhk2VNNqVL7f6S189UzShHC/mR9EE8rDdSkdpN0KflSwWY/gWjDrNxxisg8Fp1ZCD9jLMo4m0OUfeUA==}
+  '@esbuild/android-x64@0.28.1':
+    resolution: {integrity: sha512-dbwY7ltSMDWsRatcRpCnES4F+im88OCUgGZjy52shC7GqHRE/cYlxNbB4Z4UpJswpcc4Qxd2oE/ufM0p61IKng==}
     engines: {node: '>=18'}
     cpu: [x64]
     os: [android]
 
-  '@esbuild/darwin-arm64@0.28.0':
-    resolution: {integrity: sha512-0T+A9WZm+bZ84nZBtk1ckYsOvyA3x7e2Acj1KdVfV4/2tdG4fzUp91YHx+GArWLtwqp77pBXVCPn2We7Letr0Q==}
+  '@esbuild/darwin-arm64@0.28.1':
+    resolution: {integrity: sha512-TZbWkQY7kvTAXbXUT7uVACR5cMHsDiSz9z7ZKAX/RTq/WJEk3QyRr0wZpNhBDX+/0CtdqUIJlOiodQcta6tY3Q==}
     engines: {node: '>=18'}
     cpu: [arm64]
     os: [darwin]
 
-  '@esbuild/darwin-x64@0.28.0':
-    resolution: {integrity: sha512-fyzLm/DLDl/84OCfp2f/XQ4flmORsjU7VKt8HLjvIXChJoFFOIL6pLJPH4Yhd1n1gGFF9mPwtlN5Wf82DZs+LQ==}
+  '@esbuild/darwin-x64@0.28.1':
+    resolution: {integrity: sha512-zfdzgK9ACBNZLI/CyHTOx81SyNbM6YXn7rxSgX97VjyiPl9W1i4Ka4fgKECEoFCKGpvBj5qArWIGgQjOwkgskQ==}
     engines: {node: '>=18'}
     cpu: [x64]
     os: [darwin]
 
-  '@esbuild/freebsd-arm64@0.28.0':
-    resolution: {integrity: sha512-l9GeW5UZBT9k9brBYI+0WDffcRxgHQD8ShN2Ur4xWq/NFzUKm3k5lsH4PdaRgb2w7mI9u61nr2gI2mLI27Nh3Q==}
+  '@esbuild/freebsd-arm64@0.28.1':
+    resolution: {integrity: sha512-wG2EA8ENdEI0qhkSZMjfqrdY+ziCYCPMmtZjjIwOmXFjmyzEHn+UUxk5of+SYsjtfs3VpnlC7QLzSI5hY/rOAw==}
     engines: {node: '>=18'}
     cpu: [arm64]
     os: [freebsd]
 
-  '@esbuild/freebsd-x64@0.28.0':
-    resolution: {integrity: sha512-BXoQai/A0wPO6Es3yFJ7APCiKGc1tdAEOgeTNy3SsB491S3aHn4S4r3e976eUnPdU+NbdtmBuLncYir2tMU9Nw==}
+  '@esbuild/freebsd-x64@0.28.1':
+    resolution: {integrity: sha512-i7dZ9vQgnvSCzi/rYCXNgtF/U+eKZNJBzu3eTQbRgHnM7tNSizLOkRFAl3qzVc/Op/u5YkHHa4pf/3DOYHthLQ==}
     engines: {node: '>=18'}
     cpu: [x64]
     os: [freebsd]
 
-  '@esbuild/linux-arm64@0.28.0':
-    resolution: {integrity: sha512-RVyzfb3FWsGA55n6WY0MEIEPURL1FcbhFE6BffZEMEekfCzCIMtB5yyDcFnVbTnwk+CLAgTujmV/Lgvih56W+A==}
+  '@esbuild/linux-arm64@0.28.1':
+    resolution: {integrity: sha512-yHs+0uc8+nvEAfAfxrWQKK5peSNzBc4PegcMO0EJ2hT71uA7vB8Ihg2e77R2P7SG5uYjPbHlLLmve4LLLRCf0g==}
     engines: {node: '>=18'}
     cpu: [arm64]
     os: [linux]
 
-  '@esbuild/linux-arm@0.28.0':
-    resolution: {integrity: sha512-CjaaREJagqJp7iTaNQjjidaNbCKYcd4IDkzbwwxtSvjI7NZm79qiHc8HqciMddQ6CKvJT6aBd8lO9kN/ZudLlw==}
+  '@esbuild/linux-arm@0.28.1':
+    resolution: {integrity: sha512-qVXBOHQS+d5Y722GwJzJUtOLlX7km3CraOaGormF1pDtPd2C/l1SHRPgjLunLGe51Sh5YYWKMFDyV4SxgMQYTQ==}
     engines: {node: '>=18'}
     cpu: [arm]
     os: [linux]
 
-  '@esbuild/linux-ia32@0.28.0':
-    resolution: {integrity: sha512-KBnSTt1kxl9x70q+ydterVdl+Cn0H18ngRMRCEQfrbqdUuntQQ0LoMZv47uB97NljZFzY6HcfqEZ2SAyIUTQBQ==}
+  '@esbuild/linux-ia32@0.28.1':
+    resolution: {integrity: sha512-d1z4ZuP0ajrfz/FhGT4vv278rX8KnPPJx8i5+AtK7TYbx9Le9F1hyzurZpkEyjkGa9dUGhQow4C1NmeGvqxN2w==}
     engines: {node: '>=18'}
     cpu: [ia32]
     os: [linux]
 
-  '@esbuild/linux-loong64@0.28.0':
-    resolution: {integrity: sha512-zpSlUce1mnxzgBADvxKXX5sl8aYQHo2ezvMNI8I0lbblJtp8V4odlm3Yzlj7gPyt3T8ReksE6bK+pT3WD+aJRg==}
+  '@esbuild/linux-loong64@0.28.1':
+    resolution: {integrity: sha512-M5sRjUVZrkm1OAPR3dlOYzNmN+loZKGVi1VUQGrwuqLcbR6qeAz+famMhjASeH3YVKvZz+zT1jlh/keC3Rj/lg==}
     engines: {node: '>=18'}
     cpu: [loong64]
     os: [linux]
 
-  '@esbuild/linux-mips64el@0.28.0':
-    resolution: {integrity: sha512-2jIfP6mmjkdmeTlsX/9vmdmhBmKADrWqN7zcdtHIeNSCH1SqIoNI63cYsjQR8J+wGa4Y5izRcSHSm8K3QWmk3w==}
+  '@esbuild/linux-mips64el@0.28.1':
+    resolution: {integrity: sha512-mRObBZeHh2OxcBFPWE/FjylkRgZdYuiTR3vaTozquCGOH14iP9oN4x4Ge81CoIDYQrXmIxpFumJBu5MtZpnQJQ==}
     engines: {node: '>=18'}
     cpu: [mips64el]
     os: [linux]
 
-  '@esbuild/linux-ppc64@0.28.0':
-    resolution: {integrity: sha512-bc0FE9wWeC0WBm49IQMPSPILRocGTQt3j5KPCA8os6VprfuJ7KD+5PzESSrJ6GmPIPJK965ZJHTUlSA6GNYEhg==}
+  '@esbuild/linux-ppc64@0.28.1':
+    resolution: {integrity: sha512-slScBsMAb3GFDcdrCgLwZtPYRoH2H/youv10QiZyRjmsP48fznoveWytSgCI/R0ZcUgpc0ZhIUEx6LHts8yrfQ==}
     engines: {node: '>=18'}
     cpu: [ppc64]
     os: [linux]
 
-  '@esbuild/linux-riscv64@0.28.0':
-    resolution: {integrity: sha512-SQPZOwoTTT/HXFXQJG/vBX8sOFagGqvZyXcgLA3NhIqcBv1BJU1d46c0rGcrij2B56Z2rNiSLaZOYW5cUk7yLQ==}
+  '@esbuild/linux-riscv64@0.28.1':
+    resolution: {integrity: sha512-kw0owk1o0GFETUJyW0jc0G4Yzs0BHZn0JDZ8JRT088vjJYX777BAs1fDGxAC+q831qOs2DTC96mNsG2opdfyyQ==}
     engines: {node: '>=18'}
     cpu: [riscv64]
     os: [linux]
 
-  '@esbuild/linux-s390x@0.28.0':
-    resolution: {integrity: sha512-SCfR0HN8CEEjnYnySJTd2cw0k9OHB/YFzt5zgJEwa+wL/T/raGWYMBqwDNAC6dqFKmJYZoQBRfHjgwLHGSrn3Q==}
+  '@esbuild/linux-s390x@0.28.1':
+    resolution: {integrity: sha512-/lAIjX8aYFRByhh6L5rYtPEDRqa9de/4V/juOXcta5frjvzXO4/sqEtyytse0g3zZFuWu5cDN0MkLz2qRDD2Ag==}
     engines: {node: '>=18'}
     cpu: [s390x]
     os: [linux]
 
-  '@esbuild/linux-x64@0.28.0':
-    resolution: {integrity: sha512-us0dSb9iFxIi8srnpl931Nvs65it/Jd2a2K3qs7fz2WfGPHqzfzZTfec7oxZJRNPXPnNYZtanmRc4AL/JwVzHQ==}
+  '@esbuild/linux-x64@0.28.1':
+    resolution: {integrity: sha512-u/anNYF2mmVOEDwLtnQ1wOr3EZ9sTNGLWrsYGYwHWzGA3Si84IOkHXlbWTD1NB+9/1lcnweYKO54uhxZydNzfA==}
     engines: {node: '>=18'}
     cpu: [x64]
     os: [linux]
 
-  '@esbuild/netbsd-arm64@0.28.0':
-    resolution: {integrity: sha512-CR/RYotgtCKwtftMwJlUU7xCVNg3lMYZ0RzTmAHSfLCXw3NtZtNpswLEj/Kkf6kEL3Gw+BpOekRX0BYCtklhUw==}
+  '@esbuild/netbsd-arm64@0.28.1':
+    resolution: {integrity: sha512-oks0DYbLwWMmaakTsCb+zL4E+aHRVLom9IJZOAthMQEPiQmydXHkziYEsGYRx0uNV/IjEKGAV941JzH02pflqw==}
     engines: {node: '>=18'}
     cpu: [arm64]
     os: [netbsd]
 
-  '@esbuild/netbsd-x64@0.28.0':
-    resolution: {integrity: sha512-nU1yhmYutL+fQ71Kxnhg8uEOdC0pwEW9entHykTgEbna2pw2dkbFSMeqjjyHZoCmt8SBkOSvV+yNmm94aUrrqw==}
+  '@esbuild/netbsd-x64@0.28.1':
+    resolution: {integrity: sha512-aeL6lAnN89Hz43Mlh1G8ARasbuoYvSITDEx0tHh5b7jJnHcssqgjy9Yx430GDpmCa6OyrKoS0aNRjKundRizGg==}
     engines: {node: '>=18'}
     cpu: [x64]
     os: [netbsd]
 
-  '@esbuild/openbsd-arm64@0.28.0':
-    resolution: {integrity: sha512-cXb5vApOsRsxsEl4mcZ1XY3D4DzcoMxR/nnc4IyqYs0rTI8ZKmW6kyyg+11Z8yvgMfAEldKzP7AdP64HnSC/6g==}
+  '@esbuild/openbsd-arm64@0.28.1':
+    resolution: {integrity: sha512-MEFJe5C3R8pwXdZ5Y21oo6m7ePiS0d9pWucn99O/wvyJZChoIQKrQDxKrGeW8F5+T0okTHesAmDeiHDTIq0V/Q==}
     engines: {node: '>=18'}
     cpu: [arm64]
     os: [openbsd]
 
-  '@esbuild/openbsd-x64@0.28.0':
-    resolution: {integrity: sha512-8wZM2qqtv9UP3mzy7HiGYNH/zjTA355mpeuA+859TyR+e+Tc08IHYpLJuMsfpDJwoLo1ikIJI8jC3GFjnRClzA==}
+  '@esbuild/openbsd-x64@0.28.1':
+    resolution: {integrity: sha512-i/ZLIOafE0Z8cI/XANJAixoJL/uRAoS2xOA3rb0xN+KK0K177cMAsQYkzHtBrtMXAKuAc7HGgcWiZ/sRC1Nxgw==}
     engines: {node: '>=18'}
     cpu: [x64]
     os: [openbsd]
 
-  '@esbuild/openharmony-arm64@0.28.0':
-    resolution: {integrity: sha512-FLGfyizszcef5C3YtoyQDACyg95+dndv79i2EekILBofh5wpCa1KuBqOWKrEHZg3zrL3t5ouE5jgr94vA+Wb2w==}
+  '@esbuild/openharmony-arm64@0.28.1':
+    resolution: {integrity: sha512-ge+Z7EXFNt2BO1oAMsVpiQ8EwndV9i1xXerAeTIK7AtPs3bKFXQM7nlRxDSIUIMeueR1CNXxqztLzdNeReKBJg==}
     engines: {node: '>=18'}
     cpu: [arm64]
     os: [openharmony]
 
-  '@esbuild/sunos-x64@0.28.0':
-    resolution: {integrity: sha512-1ZgjUoEdHZZl/YlV76TSCz9Hqj9h9YmMGAgAPYd+q4SicWNX3G5GCyx9uhQWSLcbvPW8Ni7lj4gDa1T40akdlw==}
+  '@esbuild/sunos-x64@0.28.1':
+    resolution: {integrity: sha512-BEjgtECkL3vY+SaSQ6nzVfiALUeFxpawyp8Jmf5PtYhf1Ug40N1h/hxlhts+f1FvSvarEigdxS3BlSMI2PJLcQ==}
     engines: {node: '>=18'}
     cpu: [x64]
     os: [sunos]
 
-  '@esbuild/win32-arm64@0.28.0':
-    resolution: {integrity: sha512-Q9StnDmQ/enxnpxCCLSg0oo4+34B9TdXpuyPeTedN/6+iXBJ4J+zwfQI28u/Jl40nOYAxGoNi7mFP40RUtkmUA==}
+  '@esbuild/win32-arm64@0.28.1':
+    resolution: {integrity: sha512-lCv9eK/H6ZJWbE7bh2nw54CZ9M2nupBxJcTsdk/QQnWkdSjKGuxmmH8/GWrlT1eMmZfn4dGcCjRte397WqfQXA==}
     engines: {node: '>=18'}
     cpu: [arm64]
     os: [win32]
 
-  '@esbuild/win32-ia32@0.28.0':
-    resolution: {integrity: sha512-zF3ag/gfiCe6U2iczcRzSYJKH1DCI+ByzSENHlM2FcDbEeo5Zd2C86Aq0tKUYAJJ1obRP84ymxIAksZUcdztHA==}
+  '@esbuild/win32-ia32@0.28.1':
+    resolution: {integrity: sha512-zvb/mB2bSCoJOpoCBgYKKpX6YM6mJBlBUVUtVj41DlZJVEB6/0CKlRYxP5wWl1C1ILiCoAU5wZZ4q1P3qeS6Eg==}
     engines: {node: '>=18'}
     cpu: [ia32]
     os: [win32]
 
-  '@esbuild/win32-x64@0.28.0':
-    resolution: {integrity: sha512-pEl1bO9mfAmIC+tW5btTmrKaujg3zGtUmWNdCw/xs70FBjwAL3o9OEKNHvNmnyylD6ubxUERiEhdsL0xBQ9efw==}
+  '@esbuild/win32-x64@0.28.1':
+    resolution: {integrity: sha512-bm4Mowrv+GXMlpWX++EcXw/iLyd1o3+bJkC2DkWXYVvgZCqD/bSj9ctZeAMC3cIxgjRVR2Dufaiu4YPxr5gW1A==}
     engines: {node: '>=18'}
     cpu: [x64]
     os: [win32]
@@ -5158,8 +5158,8 @@ packages:
   esast-util-from-js@2.0.1:
     resolution: {integrity: sha512-8Ja+rNJ0Lt56Pcf3TAmpBZjmx8ZcK5Ts4cAzIOjsjevg9oSXJnl6SUQ2EevU8tv3h6ZLWmoKL5H4fgWvdvfETw==}
 
-  esbuild@0.28.0:
-    resolution: {integrity: sha512-sNR9MHpXSUV/XB4zmsFKN+QgVG82Cc7+/aaxJ8Adi8hyOac+EXptIp45QBPaVyX3N70664wRbTcLTOemCAnyqw==}
+  esbuild@0.28.1:
+    resolution: {integrity: sha512-HrJrvZv5ayxBzPfwphOoNzkzOIIlifzk0KJrGK2c8R4+LKpMtpYLQeUdjnwjWv/LZlkH2laZk+4w78pi99D4Vw==}
     engines: {node: '>=18'}
     hasBin: true
 
@@ -8366,7 +8366,7 @@ snapshots:
 
   '@copilotkit/aimock@1.27.3(vitest@4.1.7)':
     optionalDependencies:
-      vitest: 4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.1)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))
+      vitest: 4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.1)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))
 
   '@create-markdown/preview@2.0.3(shiki@4.1.0)':
     optionalDependencies:
@@ -8489,82 +8489,82 @@ snapshots:
       tslib: 2.8.1
     optional: true
 
-  '@esbuild/aix-ppc64@0.28.0':
+  '@esbuild/aix-ppc64@0.28.1':
     optional: true
 
-  '@esbuild/android-arm64@0.28.0':
+  '@esbuild/android-arm64@0.28.1':
     optional: true
 
-  '@esbuild/android-arm@0.28.0':
+  '@esbuild/android-arm@0.28.1':
     optional: true
 
-  '@esbuild/android-x64@0.28.0':
+  '@esbuild/android-x64@0.28.1':
     optional: true
 
-  '@esbuild/darwin-arm64@0.28.0':
+  '@esbuild/darwin-arm64@0.28.1':
     optional: true
 
-  '@esbuild/darwin-x64@0.28.0':
+  '@esbuild/darwin-x64@0.28.1':
     optional: true
 
-  '@esbuild/freebsd-arm64@0.28.0':
+  '@esbuild/freebsd-arm64@0.28.1':
     optional: true
 
-  '@esbuild/freebsd-x64@0.28.0':
+  '@esbuild/freebsd-x64@0.28.1':
     optional: true
 
-  '@esbuild/linux-arm64@0.28.0':
+  '@esbuild/linux-arm64@0.28.1':
     optional: true
 
-  '@esbuild/linux-arm@0.28.0':
+  '@esbuild/linux-arm@0.28.1':
     optional: true
 
-  '@esbuild/linux-ia32@0.28.0':
+  '@esbuild/linux-ia32@0.28.1':
     optional: true
 
-  '@esbuild/linux-loong64@0.28.0':
+  '@esbuild/linux-loong64@0.28.1':
     optional: true
 
-  '@esbuild/linux-mips64el@0.28.0':
+  '@esbuild/linux-mips64el@0.28.1':
     optional: true
 
-  '@esbuild/linux-ppc64@0.28.0':
+  '@esbuild/linux-ppc64@0.28.1':
     optional: true
 
-  '@esbuild/linux-riscv64@0.28.0':
+  '@esbuild/linux-riscv64@0.28.1':
     optional: true
 
-  '@esbuild/linux-s390x@0.28.0':
+  '@esbuild/linux-s390x@0.28.1':
     optional: true
 
-  '@esbuild/linux-x64@0.28.0':
+  '@esbuild/linux-x64@0.28.1':
     optional: true
 
-  '@esbuild/netbsd-arm64@0.28.0':
+  '@esbuild/netbsd-arm64@0.28.1':
     optional: true
 
-  '@esbuild/netbsd-x64@0.28.0':
+  '@esbuild/netbsd-x64@0.28.1':
     optional: true
 
-  '@esbuild/openbsd-arm64@0.28.0':
+  '@esbuild/openbsd-arm64@0.28.1':
     optional: true
 
-  '@esbuild/openbsd-x64@0.28.0':
+  '@esbuild/openbsd-x64@0.28.1':
     optional: true
 
-  '@esbuild/openharmony-arm64@0.28.0':
+  '@esbuild/openharmony-arm64@0.28.1':
     optional: true
 
-  '@esbuild/sunos-x64@0.28.0':
+  '@esbuild/sunos-x64@0.28.1':
     optional: true
 
-  '@esbuild/win32-arm64@0.28.0':
+  '@esbuild/win32-arm64@0.28.1':
     optional: true
 
-  '@esbuild/win32-ia32@0.28.0':
+  '@esbuild/win32-ia32@0.28.1':
     optional: true
 
-  '@esbuild/win32-x64@0.28.0':
+  '@esbuild/win32-x64@0.28.1':
     optional: true
 
   '@eshaz/web-worker@1.2.2': {}
@@ -10192,13 +10192,13 @@ snapshots:
 
   '@urbit/aura@3.0.0': {}
 
-  '@vitest/browser-playwright@4.1.7(playwright@1.60.0)(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))(vitest@4.1.7)':
+  '@vitest/browser-playwright@4.1.7(playwright@1.60.0)(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))(vitest@4.1.7)':
     dependencies:
-      '@vitest/browser': 4.1.7(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))(vitest@4.1.7)
-      '@vitest/mocker': 4.1.7(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))
+      '@vitest/browser': 4.1.7(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))(vitest@4.1.7)
+      '@vitest/mocker': 4.1.7(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))
       playwright: 1.60.0
       tinyrainbow: 3.1.0
-      vitest: 4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.1)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))
+      vitest: 4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.1)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))
     transitivePeerDependencies:
       - bufferutil
       - msw
@@ -10206,29 +10206,29 @@ snapshots:
       - vite
     optional: true
 
-  '@vitest/browser-playwright@4.1.7(playwright@1.60.0)(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))(vitest@4.1.7)':
+  '@vitest/browser-playwright@4.1.7(playwright@1.60.0)(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))(vitest@4.1.7)':
     dependencies:
-      '@vitest/browser': 4.1.7(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))(vitest@4.1.7)
-      '@vitest/mocker': 4.1.7(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))
+      '@vitest/browser': 4.1.7(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))(vitest@4.1.7)
+      '@vitest/mocker': 4.1.7(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))
       playwright: 1.60.0
       tinyrainbow: 3.1.0
-      vitest: 4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.2)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))
+      vitest: 4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.2)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))
     transitivePeerDependencies:
       - bufferutil
       - msw
       - utf-8-validate
       - vite
 
-  '@vitest/browser@4.1.7(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))(vitest@4.1.7)':
+  '@vitest/browser@4.1.7(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))(vitest@4.1.7)':
     dependencies:
       '@blazediff/core': 1.9.1
-      '@vitest/mocker': 4.1.7(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))
+      '@vitest/mocker': 4.1.7(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))
       '@vitest/utils': 4.1.7
       magic-string: 0.30.21
       pngjs: 7.0.0
       sirv: 3.0.2
       tinyrainbow: 3.1.0
-      vitest: 4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.1)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))
+      vitest: 4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.1)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))
       ws: 8.21.0
     transitivePeerDependencies:
       - bufferutil
@@ -10237,16 +10237,16 @@ snapshots:
       - vite
     optional: true
 
-  '@vitest/browser@4.1.7(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))(vitest@4.1.7)':
+  '@vitest/browser@4.1.7(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))(vitest@4.1.7)':
     dependencies:
       '@blazediff/core': 1.9.1
-      '@vitest/mocker': 4.1.7(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))
+      '@vitest/mocker': 4.1.7(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))
       '@vitest/utils': 4.1.7
       magic-string: 0.30.21
       pngjs: 7.0.0
       sirv: 3.0.2
       tinyrainbow: 3.1.0
-      vitest: 4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.2)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))
+      vitest: 4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.2)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))
       ws: 8.21.0
     transitivePeerDependencies:
       - bufferutil
@@ -10266,9 +10266,9 @@ snapshots:
       obug: 2.1.2
       std-env: 4.1.0
       tinyrainbow: 3.1.0
-      vitest: 4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.1)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))
+      vitest: 4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.1)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))
     optionalDependencies:
-      '@vitest/browser': 4.1.7(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))(vitest@4.1.7)
+      '@vitest/browser': 4.1.7(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))(vitest@4.1.7)
 
   '@vitest/expect@4.1.7':
     dependencies:
@@ -10279,21 +10279,21 @@ snapshots:
       chai: 6.2.2
       tinyrainbow: 3.1.0
 
-  '@vitest/mocker@4.1.7(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))':
+  '@vitest/mocker@4.1.7(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))':
     dependencies:
       '@vitest/spy': 4.1.7
       estree-walker: 3.0.3
       magic-string: 0.30.21
     optionalDependencies:
-      vite: 8.0.14(@types/node@25.9.1)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0)
+      vite: 8.0.14(@types/node@25.9.1)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0)
 
-  '@vitest/mocker@4.1.7(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))':
+  '@vitest/mocker@4.1.7(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))':
     dependencies:
       '@vitest/spy': 4.1.7
       estree-walker: 3.0.3
       magic-string: 0.30.21
     optionalDependencies:
-      vite: 8.0.14(@types/node@25.9.2)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0)
+      vite: 8.0.14(@types/node@25.9.2)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0)
 
   '@vitest/pretty-format@4.1.7':
     dependencies:
@@ -10975,34 +10975,34 @@ snapshots:
       esast-util-from-estree: 2.0.0
       vfile-message: 4.0.3
 
-  esbuild@0.28.0:
+  esbuild@0.28.1:
     optionalDependencies:
-      '@esbuild/aix-ppc64': 0.28.0
-      '@esbuild/android-arm': 0.28.0
-      '@esbuild/android-arm64': 0.28.0
-      '@esbuild/android-x64': 0.28.0
-      '@esbuild/darwin-arm64': 0.28.0
-      '@esbuild/darwin-x64': 0.28.0
-      '@esbuild/freebsd-arm64': 0.28.0
-      '@esbuild/freebsd-x64': 0.28.0
-      '@esbuild/linux-arm': 0.28.0
-      '@esbuild/linux-arm64': 0.28.0
-      '@esbuild/linux-ia32': 0.28.0
-      '@esbuild/linux-loong64': 0.28.0
-      '@esbuild/linux-mips64el': 0.28.0
-      '@esbuild/linux-ppc64': 0.28.0
-      '@esbuild/linux-riscv64': 0.28.0
-      '@esbuild/linux-s390x': 0.28.0
-      '@esbuild/linux-x64': 0.28.0
-      '@esbuild/netbsd-arm64': 0.28.0
-      '@esbuild/netbsd-x64': 0.28.0
-      '@esbuild/openbsd-arm64': 0.28.0
-      '@esbuild/openbsd-x64': 0.28.0
-      '@esbuild/openharmony-arm64': 0.28.0
-      '@esbuild/sunos-x64': 0.28.0
-      '@esbuild/win32-arm64': 0.28.0
-      '@esbuild/win32-ia32': 0.28.0
-      '@esbuild/win32-x64': 0.28.0
+      '@esbuild/aix-ppc64': 0.28.1
+      '@esbuild/android-arm': 0.28.1
+      '@esbuild/android-arm64': 0.28.1
+      '@esbuild/android-x64': 0.28.1
+      '@esbuild/darwin-arm64': 0.28.1
+      '@esbuild/darwin-x64': 0.28.1
+      '@esbuild/freebsd-arm64': 0.28.1
+      '@esbuild/freebsd-x64': 0.28.1
+      '@esbuild/linux-arm': 0.28.1
+      '@esbuild/linux-arm64': 0.28.1
+      '@esbuild/linux-ia32': 0.28.1
+      '@esbuild/linux-loong64': 0.28.1
+      '@esbuild/linux-mips64el': 0.28.1
+      '@esbuild/linux-ppc64': 0.28.1
+      '@esbuild/linux-riscv64': 0.28.1
+      '@esbuild/linux-s390x': 0.28.1
+      '@esbuild/linux-x64': 0.28.1
+      '@esbuild/netbsd-arm64': 0.28.1
+      '@esbuild/netbsd-x64': 0.28.1
+      '@esbuild/openbsd-arm64': 0.28.1
+      '@esbuild/openbsd-x64': 0.28.1
+      '@esbuild/openharmony-arm64': 0.28.1
+      '@esbuild/sunos-x64': 0.28.1
+      '@esbuild/win32-arm64': 0.28.1
+      '@esbuild/win32-ia32': 0.28.1
+      '@esbuild/win32-x64': 0.28.1
 
   escalade@3.2.0: {}
 
@@ -13655,13 +13655,13 @@ snapshots:
 
   tsx@4.22.3:
     dependencies:
-      esbuild: 0.28.0
+      esbuild: 0.28.1
     optionalDependencies:
       fsevents: 2.3.3
 
   tsx@4.22.4:
     dependencies:
-      esbuild: 0.28.0
+      esbuild: 0.28.1
     optionalDependencies:
       fsevents: 2.3.3
 
@@ -13772,7 +13772,7 @@ snapshots:
       '@types/unist': 3.0.3
       vfile-message: 4.0.3
 
-  vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0):
+  vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0):
     dependencies:
       lightningcss: 1.32.0
       picomatch: 4.0.4
@@ -13781,13 +13781,13 @@ snapshots:
       tinyglobby: 0.2.17
     optionalDependencies:
       '@types/node': 25.9.1
-      esbuild: 0.28.0
+      esbuild: 0.28.1
       fsevents: 2.3.3
       jiti: 2.7.0
       tsx: 4.22.3
       yaml: 2.9.0
 
-  vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0):
+  vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0):
     dependencies:
       lightningcss: 1.32.0
       picomatch: 4.0.4
@@ -13796,16 +13796,16 @@ snapshots:
       tinyglobby: 0.2.17
     optionalDependencies:
       '@types/node': 25.9.2
-      esbuild: 0.28.0
+      esbuild: 0.28.1
       fsevents: 2.3.3
       jiti: 2.7.0
       tsx: 4.22.4
       yaml: 2.9.0
 
-  vitest@4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.1)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0)):
+  vitest@4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.1)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0)):
     dependencies:
       '@vitest/expect': 4.1.7
-      '@vitest/mocker': 4.1.7(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))
+      '@vitest/mocker': 4.1.7(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))
       '@vitest/pretty-format': 4.1.7
       '@vitest/runner': 4.1.7
       '@vitest/snapshot': 4.1.7
@@ -13822,21 +13822,21 @@ snapshots:
       tinyexec: 1.2.4
       tinyglobby: 0.2.17
       tinyrainbow: 3.1.0
-      vite: 8.0.14(@types/node@25.9.1)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0)
+      vite: 8.0.14(@types/node@25.9.1)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0)
       why-is-node-running: 2.3.0
     optionalDependencies:
       '@opentelemetry/api': 1.9.1
       '@types/node': 25.9.1
-      '@vitest/browser-playwright': 4.1.7(playwright@1.60.0)(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))(vitest@4.1.7)
+      '@vitest/browser-playwright': 4.1.7(playwright@1.60.0)(vite@8.0.14(@types/node@25.9.1)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.3)(yaml@2.9.0))(vitest@4.1.7)
       '@vitest/coverage-v8': 4.1.7(@vitest/browser@4.1.7)(vitest@4.1.7)
       jsdom: 29.1.1(@noble/hashes@2.0.1)
     transitivePeerDependencies:
       - msw
 
-  vitest@4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.2)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0)):
+  vitest@4.1.7(@opentelemetry/api@1.9.1)(@types/node@25.9.2)(@vitest/browser-playwright@4.1.7)(@vitest/coverage-v8@4.1.7)(jsdom@29.1.1(@noble/hashes@2.0.1))(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0)):
     dependencies:
       '@vitest/expect': 4.1.7
-      '@vitest/mocker': 4.1.7(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))
+      '@vitest/mocker': 4.1.7(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))
       '@vitest/pretty-format': 4.1.7
       '@vitest/runner': 4.1.7
       '@vitest/snapshot': 4.1.7
@@ -13853,12 +13853,12 @@ snapshots:
       tinyexec: 1.2.4
       tinyglobby: 0.2.17
       tinyrainbow: 3.1.0
-      vite: 8.0.14(@types/node@25.9.2)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0)
+      vite: 8.0.14(@types/node@25.9.2)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0)
       why-is-node-running: 2.3.0
     optionalDependencies:
       '@opentelemetry/api': 1.9.1
       '@types/node': 25.9.2
-      '@vitest/browser-playwright': 4.1.7(playwright@1.60.0)(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.0)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))(vitest@4.1.7)
+      '@vitest/browser-playwright': 4.1.7(playwright@1.60.0)(vite@8.0.14(@types/node@25.9.2)(esbuild@0.28.1)(jiti@2.7.0)(tsx@4.22.4)(yaml@2.9.0))(vitest@4.1.7)
       '@vitest/coverage-v8': 4.1.7(@vitest/browser@4.1.7)(vitest@4.1.7)
       jsdom: 29.1.1(@noble/hashes@2.0.1)
     transitivePeerDependencies:

pnpm-workspace.yaml

@@ -56,6 +56,33 @@ minimumReleaseAgeExclude:
   - "oxfmt"
   - "axios@1.15.0"
   - "discord-api-types"
+  - "esbuild@0.28.1"
+  - "@esbuild/aix-ppc64@0.28.1"
+  - "@esbuild/android-arm@0.28.1"
+  - "@esbuild/android-arm64@0.28.1"
+  - "@esbuild/android-x64@0.28.1"
+  - "@esbuild/darwin-arm64@0.28.1"
+  - "@esbuild/darwin-x64@0.28.1"
+  - "@esbuild/freebsd-arm64@0.28.1"
+  - "@esbuild/freebsd-x64@0.28.1"
+  - "@esbuild/linux-arm@0.28.1"
+  - "@esbuild/linux-arm64@0.28.1"
+  - "@esbuild/linux-ia32@0.28.1"
+  - "@esbuild/linux-loong64@0.28.1"
+  - "@esbuild/linux-mips64el@0.28.1"
+  - "@esbuild/linux-ppc64@0.28.1"
+  - "@esbuild/linux-riscv64@0.28.1"
+  - "@esbuild/linux-s390x@0.28.1"
+  - "@esbuild/linux-x64@0.28.1"
+  - "@esbuild/netbsd-arm64@0.28.1"
+  - "@esbuild/netbsd-x64@0.28.1"
+  - "@esbuild/openbsd-arm64@0.28.1"
+  - "@esbuild/openbsd-x64@0.28.1"
+  - "@esbuild/openharmony-arm64@0.28.1"
+  - "@esbuild/sunos-x64@0.28.1"
+  - "@esbuild/win32-arm64@0.28.1"
+  - "@esbuild/win32-ia32@0.28.1"
+  - "@esbuild/win32-x64@0.28.1"
   - "rolldown"
   - "sqlite-vec"
   - "sqlite-vec-*"