fix(memory): reject prompt-like memory stores by eleqtrizit · Pull Request #87142 · openclaw/openclaw

eleqtrizit · 2026-05-27T03:36:37Z

Summary

Hardens the memory LanceDB plugin so explicit memory_store calls reject prompt-like instruction payloads before they reach embeddings or persistent storage.

Changes

Reuses the existing looksLikePromptInjection(text) heuristic in the explicit memory_store tool path.
Returns a rejected tool result with reason: "prompt_injection_detected" before embedding, duplicate search, or LanceDB writes.
Adds regression coverage for the rejected path and for an ordinary preference memory still storing normally.

Validation

git diff --check upstream/main...HEAD
node --import tsx --input-type=module - with an ad hoc real-behavior script that imports the actual memory-lancedb plugin, registers the real tool, executes memory_store, uses a temp LanceDB directory, and serves a local OpenAI-compatible /embeddings endpoint with no secrets.
node scripts/run-vitest.mjs extensions/memory-lancedb/index.test.ts
node scripts/run-tsgo.mjs -p test/tsconfig/tsconfig.core.test.json --incremental --tsBuildInfoFile .artifacts/tsgo-cache/core-test.tsbuildinfo
node scripts/run-tsgo.mjs -p test/tsconfig/tsconfig.extensions.test.json --incremental --tsBuildInfoFile .artifacts/tsgo-cache/extensions-test.tsbuildinfo

Notes

The branch was rebased onto current upstream/main before this proof. The earlier check-test-types CI failure was in extensions/memory-core/src/dreaming.test.ts, outside this PR's two-file diff; after the rebase, both local core and extension test-type wrapper commands completed successfully.

Real behavior proof

Behavior addressed: explicit memory_store calls that look like prompt instructions are rejected before embedding or persistent storage, while normal preference memories still store.
Real environment tested: local source checkout on Node v22.22.2, PR head 304024bd79, actual extensions/memory-lancedb/index.ts plugin imported through node --import tsx, actual plugin register() path and memory_store.execute() tool handler, temp LanceDB database directory, and a local OpenAI-compatible /embeddings HTTP endpoint using a redacted dummy key.
Exact steps or command run after this patch: ran node --import tsx --input-type=module - with an ad hoc script that registered the actual plugin API, executed a rejected prompt-like store, checked that the embedding endpoint received zero requests and LanceDB was not created, then executed a normal preference store and opened the real LanceDB table to count stored rows.
Evidence after fix: sanitized proof output was ok: true; registered tools were memory_forget, memory_recall, and memory_store; rejected details were { action: "rejected", reason: "prompt_injection_detected" }; requestsAfterReject was 0; dbExistsAfterReject was false; stored details were { action: "created", idPresent: true }; embeddingRequestsAfterStore was 1; storedRequestInput was The user prefers concise replies; lanceDbRowCount was 1; lanceDbStoredTexts contained The user prefers concise replies.
Observed result after fix: the prompt-like payload returned a rejected tool result without touching embeddings or LanceDB, and the clean preference payload created one real LanceDB memory row through the same tool path.
What was not tested: live OpenAI API credentials, a live model being induced through web_fetch, packaged install smoke, and the full repository suite.

clawsweeper · 2026-05-27T03:44:28Z

Codex review: needs maintainer review before merge. Reviewed May 27, 2026, 12:31 AM ET / 04:31 UTC.

Summary
Adds an early looksLikePromptInjection rejection path to the memory-lancedb memory_store tool and regression coverage for rejected prompt-like text plus normal preference storage.

PR surface: Source +15, Tests +97. Total +112 across 2 files.

Reproducibility: yes. source-reproducible: current main reaches embeddings.embed(text) in memory_store without the existing prompt-injection guard. I did not execute the path locally because this review is read-only, but the code path is direct and covered by the PR's after-fix proof.

Review metrics: 1 noteworthy metric.

Memory tool behavior: 1 existing tool changed. memory_store now rejects some explicit store requests before embedding or storage, which maintainers should accept as a compatibility-sensitive behavior change before merge.

Merge readiness
Overall: 🐚 platinum hermit
Proof: 🦞 diamond lobster
Patch quality: 🐚 platinum hermit
Result: ready for maintainer review.

Overall follows the weaker of proof and patch quality, so missing proof can cap an otherwise strong patch.

Rank-up moves:

Get explicit maintainer acceptance of the fail-closed memory_store behavior before merge.

Risk before merge

Merging intentionally makes the existing memory_store tool fail closed for text matched by the shared prompt-injection heuristic, so workflows that intentionally store instruction-like notes could see false-positive rejections.

Maintainer options:

Accept fail-closed hardening (recommended)
Merge once maintainers explicitly accept that prompt-like explicit memories should be rejected by the existing heuristic despite possible false positives.
Narrow the rejection path first
If the compatibility cost is too high, adjust the heuristic or add a maintainer-approved explicit escape path before merge.
Pause for memory policy direction
If explicit memory stores should be allowed to persist instruction-like text, pause this PR and settle that broader memory safety policy first.

Next step before merge
Human handling is needed because the PR has the protected maintainer label and the remaining decision is accepting or narrowing the compatibility-sensitive fail-closed behavior.

Security
Cleared: No security or supply-chain regression found; the diff adds no dependencies, workflows, secrets handling, downloads, or package-resolution changes and reduces one prompt-injection persistence path.

Review details

Best possible solution:

Merge after maintainers explicitly accept the false-positive tradeoff and required checks remain green; otherwise narrow the heuristic or add an approved escape path before merge.

Do we have a high-confidence way to reproduce the issue?

Yes, source-reproducible: current main reaches embeddings.embed(text) in memory_store without the existing prompt-injection guard. I did not execute the path locally because this review is read-only, but the code path is direct and covered by the PR's after-fix proof.

Is this the best way to solve the issue?

Yes for the code shape: reusing the existing plugin-local heuristic before embeddings and writes is the narrowest implementation. The remaining question is maintainer acceptance of the compatibility tradeoff, not a concrete patch defect.

AGENTS.md: found and applied where relevant.

Codex review notes: model gpt-5.5, reasoning high; reviewed against 1c8a11265ba4.

Label changes

Label justifications:

P2: This is a bounded memory plugin hardening fix with limited blast radius, but it changes an existing persisted-memory tool path.
merge-risk: 🚨 compatibility: The PR changes existing memory_store behavior by rejecting prompt-like explicit store requests that current main would embed and persist.
rating: 🐚 platinum hermit: Overall readiness is 🐚 platinum hermit; proof is 🦞 diamond lobster and patch quality is 🐚 platinum hermit.
status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR. Sufficient (live_output): The PR body includes sufficient after-fix live output from actual plugin registration and tool execution using temp LanceDB storage and a local embeddings endpoint.
proof: sufficient: Contributor real behavior proof is sufficient. The PR body includes sufficient after-fix live output from actual plugin registration and tool execution using temp LanceDB storage and a local embeddings endpoint.

Evidence reviewed

PR surface:

Source +15, Tests +97. Total +112 across 2 files.

View PR surface stats

Area	Files	Added	Net
Source	1	15	+15
Tests	1	97	+97
Docs	0	0	0
Config	0	0	0
Generated	0	0	0
Other	0	0	0
Total	2	112	+112

What I checked:

Repository policy read: Read the full root AGENTS.md and the scoped extensions/AGENTS.md; the review applied the repository rule that plugin behavior changes and fail-closed paths are compatibility-sensitive. (AGENTS.md:17, 1c8a11265ba4)
Current main behavior: On current main, memory_store extracts the text parameter and immediately calls embeddings.embed(text) before duplicate search and storage, with no prompt-injection guard in the explicit tool path. (extensions/memory-lancedb/index.ts:785, 1c8a11265ba4)
Existing heuristic and sibling behavior: Current main already defines PROMPT_INJECTION_PATTERNS / looksLikePromptInjection and uses that helper to skip auto-capture, so the PR reuses an existing plugin-local policy rather than adding a parallel detector. (extensions/memory-lancedb/index.ts:520, 1c8a11265ba4)
PR implementation: The PR adds an early rejected tool result with details.action = "rejected" and reason = "prompt_injection_detected" before embedding, duplicate search, or LanceDB writes. (extensions/memory-lancedb/index.ts:782, 304024bd7990)
Regression coverage: The added test verifies a prompt-like memory_store call does not call embeddings, LanceDB loading, or add, then verifies a normal preference still stores through the same tool path. (extensions/memory-lancedb/index.test.ts:2385, 304024bd7990)
Documented tool surface: The public docs describe memory_store as the agent tool for saving important facts, preferences, decisions, and entities; rejecting prompt-like instruction payloads changes this existing user-visible tool behavior rather than adding a new tool. Public docs: docs/plugins/memory-lancedb.md. (docs/plugins/memory-lancedb.md:254, 1c8a11265ba4)

Likely related people:

steipete: Recent path history and local blame connect this person to the current memory-lancedb implementation, the existing prompt-injection helper area, embedding runtime refactors, and nearby tests. (role: recent area contributor; confidence: high; commits: 77d9ac30bb8d, eabae023ebb2, 53032505bd95; files: extensions/memory-lancedb/index.ts, extensions/memory-lancedb/index.test.ts)
brokemac79: Recent merged commits on the same plugin added public memory artifacts and preserved sidecar capability hooks, adjacent to the runtime registration surface this PR changes. (role: recent memory-lancedb contributor; confidence: medium; commits: aac1abeafff4, e6288cab9a45; files: extensions/memory-lancedb/index.ts, extensions/memory-lancedb/index.test.ts)
shakkernerd: Recent history shows memory-lancedb hook test work and merge ownership for the CJK auto-capture trigger change, which is adjacent to the shared capture/filtering policy. (role: adjacent test and trigger contributor; confidence: medium; commits: 7fa771b665d2, 4d2e70872640; files: extensions/memory-lancedb/index.ts, extensions/memory-lancedb/index.test.ts)

What the crustacean ranks mean

🦀 challenger crab: rare, exceptional readiness with strong proof, clean implementation, and convincing validation.
🦞 diamond lobster: very strong readiness with only minor maintainer review expected.
🐚 platinum hermit: good normal PR, likely mergeable with ordinary maintainer review.
🦐 gold shrimp: useful signal, but proof or patch confidence is still limited.
🦪 silver shellfish: thin signal; proof, validation, or implementation needs work.
🧂 unranked krab: not merge-ready because proof is missing/unusable or there are serious correctness or safety concerns.
🌊 off-meta tidepool: rating does not apply to this item.

Shiny media proof means a screenshot, video, or linked artifact directly shows the changed behavior. Runtime, network, CSP, and security claims still need visible diagnostics.

How this review workflow works

ClawSweeper keeps one durable marker-backed review comment per issue or PR.
Re-runs edit this comment so the latest verdict, findings, and automation markers stay together instead of adding duplicate bot comments.
A fresh review can be triggered by eligible @clawsweeper re-review comments, exact-item GitHub events, scheduled/background review runs, or manual workflow dispatch.
PR/issue authors and users with repository write access can comment @clawsweeper re-review or @clawsweeper re-run on an open PR or issue to request a fresh review only.
Maintainers can also comment @clawsweeper review to request a fresh review only.
Fresh-review commands do not start repair, autofix, rebase, CI repair, or automerge.
Maintainer-only repair and merge flows require explicit commands such as @clawsweeper autofix, @clawsweeper automerge, @clawsweeper fix ci, or @clawsweeper address review.
Maintainers can comment @clawsweeper explain to ask for more context, or @clawsweeper stop to stop active automation.

clawsweeper · 2026-05-27T03:49:49Z

ClawSweeper PR egg

✨ Hatched: 🌱 uncommon Neon Signal Puff

Hatch command

Comment @clawsweeper hatch when this PR is hatchable.

Hatchability rules:

Merged PRs are hatchable.
Open PRs are hatchable when they are status: 👀 ready for maintainer look, status: 🚀 automerge armed, or labeled clawsweeper:automerge.
Closed unmerged PRs are hatchable only when one of those hatchable labels is still present in the durable record.

Rarity: 🌱 uncommon.
Trait: keeps receipts.
Image traits: location merge queue dock; accessory commit compass; palette pearl, teal, and neon green; mood mischievous; pose guarding a tiny green check; shell paper lantern shell; lighting clean product lighting; background miniature CI buoys.
Share on X: post this hatch
Copy: My PR egg hatched a 🌱 uncommon Neon Signal Puff in ClawSweeper.

What is this egg doing here?

Eggs appear after the PR passes real-behavior proof. It is here for vibes, not verdicts: it does not change labels, ratings, merge decisions, or automation.
The shell reacts to review momentum: open follow-up work warms it up, re-review makes it wobble, and a clean final review lets it hatch.
Hatchability usually comes from sufficient real-behavior proof, no blocking P0/P1/P2 findings, no security attention needed, and clean correctness. A merged PR is already final, so merge makes the egg hatchable independently.
The hatch is seeded from this repository and PR number, so the same PR keeps the same creature; the reviewed head SHA can only change safe visual details.
Rarity is just collectible sparkle: 🥚 common, 🌱 uncommon, 💎 rare, ✨ glimmer, and 🌈 legendary.

eleqtrizit · 2026-05-27T04:14:19Z

@clawsweeper re-review

clawsweeper · 2026-05-27T04:14:22Z

🦞🧹
ClawSweeper re-review requested.

I asked ClawSweeper to review this item again.
Action: item re-review queued (workflow sweep.yml, event repository_dispatch).
Result: the existing ClawSweeper review comment will be edited in place when the review finishes.

Re-review progress:

State: Superseded
Detail: A newer re-review for this item started before this run finished, so GitHub cancelled this older run. Check the latest ClawSweeper run for the current result.
Run: https://github.com/openclaw/clawsweeper/actions/runs/26490365653
Updated: 2026-05-27T04:25:11.289Z

* fix(memory): reject prompt-like memory stores * fix(changelog): mention memory store rejection

…026.5.26) (#682) This PR contains the following updates: | Package | Update | Change | |---|---|---| | [ghcr.io/openclaw/openclaw](https://openclaw.ai) ([source](https://github.com/openclaw/openclaw)) | patch | `2026.5.22` → `2026.5.26` | --- > ⚠️ **Warning** > > Some dependencies could not be looked up. Check the [Dependency Dashboard](issues/567) for more information. --- ### Release Notes <details> <summary>openclaw/openclaw (ghcr.io/openclaw/openclaw)</summary> ### [`v2026.5.26`](https://github.com/openclaw/openclaw/blob/HEAD/CHANGELOG.md#2026526) [Compare Source](https://github.com/openclaw/openclaw/compare/v2026.5.22...v2026.5.26) ##### Highlights - Faster Gateway and replies: startup avoids repeated plugin, channel, session, usage-cost, warning, scheduled-service, and filesystem scans; visible replies separate user-facing sends from slower follow-up work; Gateway runtime/session caches churn less under load. - Transcripts are core: transcript-backed meeting summaries, source-provider chunks, cleaned user turns, media provenance, Codex mirrors, WebChat replies, and CLI/TUI replay now use one more reliable transcript path. - More channels are production-ready: Telegram keeps typing/progress context and forum topics, iMessage handles attachment roots, remote media staging, and duplicate local Messages sources, WhatsApp restores group/media behavior, Discord improves voice playback and model picking, and Signal/iMessage/WhatsApp get reaction approvals. - Better voice and Talk: realtime Talk runs can be inspected, steered, cancelled, or followed up from Web UI and Discord voice; wake-name handling is more tolerant without letting ambient speech trigger agents. - Safer content boundaries: Browser snapshot reads honor SSRF policy, system-event text cannot spoof nested prompt markers, fetched file text is wrapped as external content, ClickClack inbound sender allowlists run before agent dispatch, stale device tokens are rejected, and serialized tool-call text is scrubbed from replies. - Providers, Codex, and local models are steadier: named auth profiles, OpenAI sampling params, Codex app-server resume/timeout/usage-limit recovery, dynamic tool-schema guards, xAI usage-limit surfacing, Ollama top-p normalization, and local approval resolution reduce provider-specific dead ends. - More reliable install/update/release paths: Alpine installs, trusted runtime fallback roots, stable update channels, Docker/package timeouts, Windows Scheduled Tasks, Windows/macOS proof lanes, Testbox/Crabbox delegation, plugin publish checks, and macOS runner bootstraps all got hardened. - Better observability: Activity tab, gateway secret-prep traces, tool/model stream progress, explicit fast-mode status, systemd Gateway hygiene, OpenTelemetry LLM spans, release performance evidence, and richer telemetry signals make failures easier to inspect. ##### Changes - Transcripts: add core transcript capture and source-provider support for transcript-backed meeting summaries, including the renamed Transcripts docs, CLI surface, source-provider chunks, and cleaned user-turn persistence. - Auth: add named model login profiles and supported credential migration for Hermes, OpenCode, and Codex auth profiles, with explicit opt-out and non-interactive controls. ([#85667](https://github.com/openclaw/openclaw/issues/85667)) Thanks [@fuller-stack-dev](https://github.com/fuller-stack-dev). - Diagnostics: trace gateway secret preparation, classify skill/tool usage, surface model stream progress, add OpenTelemetry LLM content spans, and expose alertable telemetry for blocked tools, failover, stale sessions, liveness, oversized payloads, and webhook ingress. ([#83019](https://github.com/openclaw/openclaw/issues/83019), [#80370](https://github.com/openclaw/openclaw/issues/80370), [#86191](https://github.com/openclaw/openclaw/issues/86191)) - Channels: add Signal reaction approvals, iMessage thumb approval reactions, and WhatsApp thumb approval reaction support so mobile approval flows work without textual `/approve` commands. ([#85894](https://github.com/openclaw/openclaw/issues/85894), [#85952](https://github.com/openclaw/openclaw/issues/85952), [#85477](https://github.com/openclaw/openclaw/issues/85477)) - Agents/API: forward OpenAI sampling params through the Gateway and expose estimated context-budget status for active agent runs. ([#84094](https://github.com/openclaw/openclaw/issues/84094)) - TUI/status: queue prompts submitted while an agent is busy and show explicit fast-mode state plus richer systemd Gateway hygiene in status output. ([#86722](https://github.com/openclaw/openclaw/issues/86722), [#87115](https://github.com/openclaw/openclaw/issues/87115), [#86976](https://github.com/openclaw/openclaw/issues/86976)) - Exec approvals: hide durable approval actions that are unavailable for the current prompt and keep approval runtime tokens local-only so stale prompts cannot offer misleading controls. ([#86270](https://github.com/openclaw/openclaw/issues/86270), [#86359](https://github.com/openclaw/openclaw/issues/86359)) - Plugin SDK: add reaction approval helpers and keep diagnostic event root exports discoverable across function-name and alias-bound module graphs. ([#86735](https://github.com/openclaw/openclaw/issues/86735), [#87084](https://github.com/openclaw/openclaw/issues/87084)) - Android/iOS: add the Android pair-new-gateway action and improve mobile Talk mode surfaces, including iOS realtime Talk mode and Android offline voice/gateway recovery. ([#86798](https://github.com/openclaw/openclaw/issues/86798), [#86355](https://github.com/openclaw/openclaw/issues/86355)) Thanks [@ngutman](https://github.com/ngutman). - Performance: cache plugin metadata snapshots, package realpaths, stable gateway metadata, model cost indexes, channel resolution, usage-cost indexes, and session/auth hot-path facts so common Gateway and reply paths do less rediscovery. ([#84649](https://github.com/openclaw/openclaw/issues/84649), [#85843](https://github.com/openclaw/openclaw/issues/85843), [#86517](https://github.com/openclaw/openclaw/issues/86517), [#86678](https://github.com/openclaw/openclaw/issues/86678)) - Voice: expose shared realtime turn-context tracking through the realtime voice SDK and reuse it for Discord speaker attribution and wake-name context recovery. - Voice: reuse shared realtime output activity tracking in Google Meet command and node audio bridges, including recent-output checks for local barge-in detection. - Voice: expose shared realtime output activity tracking through the realtime voice SDK and reuse it for Discord playback activity and barge-in decisions. - Voice: expose shared realtime consult question matching, speakable-result extraction, and alias-aware forced-consult coordination through the realtime voice SDK, then reuse it in Gateway Talk, Voice Call, and Discord voice paths. - Voice: share activation-name matching and consult-transcript screening through the realtime voice SDK so Discord, browser voice, and meeting surfaces can reuse one implementation. - Cron: default `cron.maxConcurrentRuns` to 8 so scheduled automations and their isolated agent turns can make progress in parallel without explicit configuration. - QA-Lab: add `qa coverage --match <query>` so focused proof selection can discover matching scenarios from existing metadata before running live or remote lanes. - Discord/model picker: surface an alpha-bucket select (e.g. `A–G (12) · H–N (18) · O–Z (5)`) when the provider list or a provider's model list exceeds 25 items, so configs with `provider/*` wildcards stay one click from the right page instead of paginating through prev/next; falls back to numeric chunks when every item shares the same first letter. - Control UI: add an ephemeral Activity tab for sanitized live tool activity summaries without persisting raw telemetry. Fixes [#12831](https://github.com/openclaw/openclaw/issues/12831). Thanks [@BunsDev](https://github.com/BunsDev). - Build: include `ui:build` in the `full` and `ciArtifacts` profiles of `scripts/build-all.mjs` so `pnpm build` always rebuilds `dist/control-ui` after `tsdown` cleans `dist`, removing the second-command requirement and the missing-asset failure mode for source/runtime installs and CI artifact uploads. ([#85206](https://github.com/openclaw/openclaw/issues/85206)) - iOS: improve Talk mode with direct realtime voice sessions, compact toolbar status, and responsive voice waveform feedback. ([#86355](https://github.com/openclaw/openclaw/issues/86355)) Thanks [@ngutman](https://github.com/ngutman). - Media: replace the Sharp image backend with Rastermill for metadata, resizing, EXIF orientation, and PNG alpha-preserving optimization so OpenClaw no longer installs Sharp or the WhatsApp Jimp fallback for image processing. ([#86437](https://github.com/openclaw/openclaw/issues/86437)) - Codex: update the bundled Codex CLI to 0.134.0 and keep native compaction disabled for budget-triggered app-server turns so OpenClaw owns the recovery boundary. ([#86772](https://github.com/openclaw/openclaw/issues/86772)) ##### Fixes - Memory/security: reject prompt-like text submitted through the explicit `memory_store` tool before embedding or storage, matching the existing auto-capture prompt-injection filter. ([#87142](https://github.com/openclaw/openclaw/issues/87142)) - Gateway/security: enable the default auth rate limiter for remote non-browser and HTTP gateway auth failures when `gateway.auth.rateLimit` is unset, while preserving the loopback exemption. ([#87148](https://github.com/openclaw/openclaw/issues/87148)) - Prompt hardening: route untrusted group prompt metadata through sanitized untrusted structured context while preserving trusted operator-configured group system prompts and aligning the plugin SDK docs/test helpers. ([#87144](https://github.com/openclaw/openclaw/issues/87144)) - Security/content boundaries: validate Browser snapshot tab URLs against SSRF policy before ChromeMCP or direct CDP reads, sanitize queued system-event text so untrusted plugin/channel labels cannot spoof nested prompt markers, wrap fetched file text and metadata as external content, apply ClickClack `allowFrom` sender allowlists before agent dispatch, reject RPCs from invalidated device-token clients during rotation, require staged sandbox media refs, and scrub serialized tool-call text from replies. ([#78526](https://github.com/openclaw/openclaw/issues/78526), [#87094](https://github.com/openclaw/openclaw/issues/87094), [#87062](https://github.com/openclaw/openclaw/issues/87062), [#83741](https://github.com/openclaw/openclaw/issues/83741), [#70707](https://github.com/openclaw/openclaw/issues/70707), [#86924](https://github.com/openclaw/openclaw/issues/86924)) Thanks [@zsxsoft](https://github.com/zsxsoft), [@ttzero25](https://github.com/ttzero25), and [@mmaps](https://github.com/mmaps). - Transcripts/user turns: persist CLI, WebChat, media, follow-up, hook, and Codex-mirror user turns to the admitted session target; keep cleaned transcript text, inline image routing, provenance metadata, replay hooks, and fallback paths idempotent when runtimes fail or restart. - TUI/status/onboarding/UI: queue busy TUI prompts instead of dropping them, preserve the configured default model during onboarding, show failed tool results as errors, show config-open failures in Control UI, keep status JSON plugin scans healthy, preserve xAI usage-limit errors locally, and expose explicit fast-mode/systemd state. ([#86722](https://github.com/openclaw/openclaw/issues/86722), [#87000](https://github.com/openclaw/openclaw/issues/87000), [#85786](https://github.com/openclaw/openclaw/issues/85786), [#87108](https://github.com/openclaw/openclaw/issues/87108), [#87001](https://github.com/openclaw/openclaw/issues/87001), [#86614](https://github.com/openclaw/openclaw/issues/86614), [#87115](https://github.com/openclaw/openclaw/issues/87115), [#86976](https://github.com/openclaw/openclaw/issues/86976)) - Plugin commands/SDK: preserve plugin LLM command auth, bind native plugin command dispatch to the host agent's LLM auth, keep `onDiagnosticEvent` exports discoverable through `Function.name`, stabilize diagnostic event root aliases, correlate pathless read diagnostics, suppress transient runner failures in channel command paths, and repair local approval resolution. ([#85936](https://github.com/openclaw/openclaw/issues/85936), [#87084](https://github.com/openclaw/openclaw/issues/87084), [#86977](https://github.com/openclaw/openclaw/issues/86977), [#87069](https://github.com/openclaw/openclaw/issues/87069), [#86771](https://github.com/openclaw/openclaw/issues/86771)) - Codex/providers: keep WebChat delivery hints out of user prompts, avoid false queued-terminal idle timeouts, share the native hook relay registry, quarantine unsupported dynamic tool schemas, preserve Claude resumed-session system prompts, normalize greedy Ollama `top_p`, preserve per-agent thinking defaults for ingress runs, and avoid native compaction takeover on budget-triggered Codex turns. ([#87096](https://github.com/openclaw/openclaw/issues/87096), [#73950](https://github.com/openclaw/openclaw/issues/73950), [#87049](https://github.com/openclaw/openclaw/issues/87049), [#86689](https://github.com/openclaw/openclaw/issues/86689), [#86772](https://github.com/openclaw/openclaw/issues/86772)) - Gateway/perf/release: reuse startup-warning metadata and prepared auth stores, avoid cloning live-switch and lifecycle session caches on read paths, defer warning and scheduled-service fallback imports, trim Gateway session/startup/runtime CPU churn, skip duplicate turn session touches, stop chat timeout fallback cascades, drop stale subagent announce history, bound benchmark/watch/kitchen-sink teardown waits, bound macOS/package/onboarding/plugin smoke commands, bound install finalization probes, resolve Parallels npm-update commands from guest `PATH`, and bootstrap raw AWS macOS Node/pnpm commands through `/usr/bin/env`. ([#86997](https://github.com/openclaw/openclaw/issues/86997)) - Reply/perf: reduce visible reply delivery latency by preserving Telegram typing/progress context, lazy-loading slash-command startup metadata, avoiding hot-path model hydration, flag-gating Codex profiler timing, deferring context compaction maintenance, and tracking delivery timing. ([#86989](https://github.com/openclaw/openclaw/issues/86989), [#86990](https://github.com/openclaw/openclaw/issues/86990), [#86991](https://github.com/openclaw/openclaw/issues/86991), [#86992](https://github.com/openclaw/openclaw/issues/86992), [#86993](https://github.com/openclaw/openclaw/issues/86993), [#86994](https://github.com/openclaw/openclaw/issues/86994)) Thanks [@keshavbotagent](https://github.com/keshavbotagent). - Reply/source delivery: keep TUI, Control UI, media, TTS, transcript, and Codex source-reply finals live without duplicate terminal events or stale replay artifacts. - Agents/replay: repair legacy tool results before replay, preserve `sessions_spawn` transcript payloads, restore current guard checks, stage sandboxed workspace media, and keep duplicate transcripts tool display metadata from reappearing. ([#82203](https://github.com/openclaw/openclaw/issues/82203), [#86934](https://github.com/openclaw/openclaw/issues/86934), [#87025](https://github.com/openclaw/openclaw/issues/87025)) Thanks [@martingarramon](https://github.com/martingarramon), [@vincentkoc](https://github.com/vincentkoc), and [@joshavant](https://github.com/joshavant). - Agents/sessions: handle active-fallback failures in `sessions_send` so fallback routing reports the real failure and does not leave callers with an ambiguous dropped send. ([#86638](https://github.com/openclaw/openclaw/issues/86638)) - Agents/hooks/subagents: enforce default hook agent allowlists, recover failed subagent lifecycle completions, and keep node task lifecycle cleanup from closing the Gateway listener. ([#86101](https://github.com/openclaw/openclaw/issues/86101)) - Codex: project newer OpenClaw chat history into resumed app-server threads and keep Codex turn timeouts inside the Codex runtime boundary so timeouts do not poison shared app-server clients or fall through to unrelated provider fallback. ([#86677](https://github.com/openclaw/openclaw/issues/86677), [#86476](https://github.com/openclaw/openclaw/issues/86476)) Thanks [@TurboTheTurtle](https://github.com/TurboTheTurtle) and [@pashpashpash](https://github.com/pashpashpash). - Config/doctor/update: narrow profiled tool-section doctor repair, keep runtime-injected legacy web-search provider config out of user-authored config validation, and keep prerelease tags excluded from stable updater resolution. ([#87030](https://github.com/openclaw/openclaw/issues/87030), [#86818](https://github.com/openclaw/openclaw/issues/86818), [#86559](https://github.com/openclaw/openclaw/issues/86559)) Thanks [@joshavant](https://github.com/joshavant), [@luoyanglang](https://github.com/luoyanglang), and [@stevenepalmer](https://github.com/stevenepalmer). - Doctor/runtime: validate active bundled MCP tool schemas through the same runtime projection path so unsupported MCP input schemas are reported and quarantined instead of poisoning assistant startup. - CLI/Windows: add a Windows-only stack-size respawn for stack-heavy startup paths, default CLI logs to local timestamps, and validate timeout/banner TTY state more strictly. ([#87031](https://github.com/openclaw/openclaw/issues/87031), [#85387](https://github.com/openclaw/openclaw/issues/85387)) Thanks [@giodl73-repo](https://github.com/giodl73-repo) and [@vincentkoc](https://github.com/vincentkoc). - Locking/security: require owner identity proof before stale plugin lock removal, memoize session lock owner arguments, and avoid writing default exec approval stores unless policy state actually changed. ([#86814](https://github.com/openclaw/openclaw/issues/86814), [#86964](https://github.com/openclaw/openclaw/issues/86964)) Thanks [@Alix-007](https://github.com/Alix-007) and [@vincentkoc](https://github.com/vincentkoc). - Install/release: bound Docker package build, inventory, pack, and tarball preparation with process-group timeouts; pin shrinkwrap patch drift to the pnpm lock; harden macOS restart and dSYM packaging; and run release Docker/live timeout wrappers in the foreground so child processes cannot wedge gates. - QA/Telegram: bound Telegram user credential tar and broker calls so live proof setup fails with a timeout instead of waiting for the outer Crabbox job deadline. - QA/Tool Search: bound gateway E2E HTTP probes, run only the fixture plugin, and clean up temporary fixture trees after the compact tool-catalog proof completes. - Telegram/network: treat `ENETDOWN` as a transient pre-connect network failure so Telegram sends, gateway unhandled-rejection handling, and cron network retries follow the same recovery path as sibling network outages. ([#86762](https://github.com/openclaw/openclaw/issues/86762)) Thanks [@TurboTheTurtle](https://github.com/TurboTheTurtle). - Telegram: preserve inbound text entities, overlapping DM replies, account topic cache sidecars, outbound reply context, targeted bot-command mentions, durable group retry targets, forum topic names, and native progress callbacks. ([#83873](https://github.com/openclaw/openclaw/issues/83873), [#85361](https://github.com/openclaw/openclaw/issues/85361), [#85555](https://github.com/openclaw/openclaw/issues/85555), [#85656](https://github.com/openclaw/openclaw/issues/85656), [#85709](https://github.com/openclaw/openclaw/issues/85709), [#86299](https://github.com/openclaw/openclaw/issues/86299), [#86553](https://github.com/openclaw/openclaw/issues/86553)) Thanks [@SebTardif](https://github.com/SebTardif), [@luoyanglang](https://github.com/luoyanglang), and [@neeravmakwana](https://github.com/neeravmakwana). - iMessage: read image attachments from local Messages attachment roots, dedupe duplicate local Messages-source accounts, seed direct DM history, fix image/group media attachment commands, advance catchup cursors after live handling, and keep slash-command acknowledgements in the source conversation. ([#82642](https://github.com/openclaw/openclaw/issues/82642), [#85475](https://github.com/openclaw/openclaw/issues/85475), [#86569](https://github.com/openclaw/openclaw/issues/86569), [#86705](https://github.com/openclaw/openclaw/issues/86705), [#86706](https://github.com/openclaw/openclaw/issues/86706), [#86770](https://github.com/openclaw/openclaw/issues/86770)) Thanks [@homer-byte](https://github.com/homer-byte), [@TurboTheTurtle](https://github.com/TurboTheTurtle), [@swang430](https://github.com/swang430), and [@OmarShahine](https://github.com/OmarShahine). - WhatsApp/QQ/Twitch/IRC/Slack: restore WhatsApp ack identity and group-drop warnings, make QQ Bot media respect `OPENCLAW_HOME`, serialize Twitch auth disconnects, store IRC channel routes canonically, and keep Slack downloaded files out of reply media. ([#83833](https://github.com/openclaw/openclaw/issues/83833), [#85309](https://github.com/openclaw/openclaw/issues/85309), [#85777](https://github.com/openclaw/openclaw/issues/85777), [#85794](https://github.com/openclaw/openclaw/issues/85794), [#85906](https://github.com/openclaw/openclaw/issues/85906), [#86318](https://github.com/openclaw/openclaw/issues/86318), [#86697](https://github.com/openclaw/openclaw/issues/86697)) Thanks [@sliverp](https://github.com/sliverp), [@neeravmakwana](https://github.com/neeravmakwana), and [@Kailigithub](https://github.com/Kailigithub). - Discord/voice: improve voice playback and wake replies, bucket large model picker menus, merge media captions into one message, route metadata through configured proxies, restore numeric channel sends, suppress self-reply echoes, and tighten wake matching without breaking fuzzy wake phrases. ([#80227](https://github.com/openclaw/openclaw/issues/80227), [#86238](https://github.com/openclaw/openclaw/issues/86238), [#86487](https://github.com/openclaw/openclaw/issues/86487), [#86571](https://github.com/openclaw/openclaw/issues/86571), [#86595](https://github.com/openclaw/openclaw/issues/86595), [#86601](https://github.com/openclaw/openclaw/issues/86601)) - Codex: preserve native web-search metadata, keep oversized native thread reuse, bridge CLI API-key auth into the app server, preserve sandbox bootstrap path style, recover context-window prompt errors, honor yolo approval policy, disable native thread personality, and route compaction through Codex auth. ([#85378](https://github.com/openclaw/openclaw/issues/85378), [#85542](https://github.com/openclaw/openclaw/issues/85542), [#85891](https://github.com/openclaw/openclaw/issues/85891), [#85909](https://github.com/openclaw/openclaw/issues/85909), [#86408](https://github.com/openclaw/openclaw/issues/86408)) - Agents/runtime: enforce session lock max-hold reclaim, release embedded-attempt locks on all exits, treat aborted subagent runs as terminal, avoid runtime model hydration on hot paths, disclose scoped session list counts, derive overflow budgets from provider errors, and keep fallback errors scoped to the active model candidate. ([#70473](https://github.com/openclaw/openclaw/issues/70473), [#85764](https://github.com/openclaw/openclaw/issues/85764), [#86014](https://github.com/openclaw/openclaw/issues/86014), [#86134](https://github.com/openclaw/openclaw/issues/86134), [#86427](https://github.com/openclaw/openclaw/issues/86427), [#86944](https://github.com/openclaw/openclaw/issues/86944)) Thanks [@openperf](https://github.com/openperf), [@fuller-stack-dev](https://github.com/fuller-stack-dev), [@zhangguiping-xydt](https://github.com/zhangguiping-xydt), and [@ferminquant](https://github.com/ferminquant). - Config/update/doctor: retry config recovery after failed backup restore, skip shell env fallback on Windows, exclude prerelease tags from the stable git channel, support deep config edits, warn instead of aborting on unreadable cron stores, prune stale bundled plugin paths, and avoid duplicate restart prompts when the Gateway is already healthy. ([#85739](https://github.com/openclaw/openclaw/issues/85739), [#85787](https://github.com/openclaw/openclaw/issues/85787), [#86060](https://github.com/openclaw/openclaw/issues/86060), [#86260](https://github.com/openclaw/openclaw/issues/86260), [#86384](https://github.com/openclaw/openclaw/issues/86384), [#86533](https://github.com/openclaw/openclaw/issues/86533)) Thanks [@liaoyl830](https://github.com/liaoyl830). - Install/release: support Alpine CLI installs and runtime floors, prefer trusted startup argv runtime fallback roots, reject stale CLI node runtimes, avoid npm `min-release-age` installer failures, bound npm/package/Docker install phases, restore config parent ownership in Docker, seed Docker lockfile package tarballs before prune, make release/plugin prerelease checks fail closed instead of hanging or false-greening, and use host-visible Crabbox local work roots for Docker-backed proof. ([#85491](https://github.com/openclaw/openclaw/issues/85491)) - Windows daemon: keep Scheduled Task gateway launches running on battery power and avoid workgroup-machine prompts for a domain user during task installation. ([#59299](https://github.com/openclaw/openclaw/issues/59299)) - Security: avoid printing Gateway tokens in Docker, validate plugin model-pattern regexes safely, escape transcript metadata field names, harden session allowlist glob matching, audit Claude permission overrides under YOLO, and require explicit allow for ACP auto approvals. ([#85849](https://github.com/openclaw/openclaw/issues/85849), [#85934](https://github.com/openclaw/openclaw/issues/85934), [#86046](https://github.com/openclaw/openclaw/issues/86046), [#86557](https://github.com/openclaw/openclaw/issues/86557)) - Media/images: replace Sharp with Rastermill, keep EXIF normalization best-effort, normalize HEIC/HEIF before image descriptions, route Codex image API keys through OpenAI, preserve image compression metadata, and auto-scale live tool result caps. ([#85776](https://github.com/openclaw/openclaw/issues/85776), [#86037](https://github.com/openclaw/openclaw/issues/86037), [#86437](https://github.com/openclaw/openclaw/issues/86437), [#86857](https://github.com/openclaw/openclaw/issues/86857), [#86923](https://github.com/openclaw/openclaw/issues/86923)) - Memory: prevent semantic vector indexes from silently degrading when embeddings are unavailable, stop doctor OOMs on large session stores, preserve sidecar hooks/artifacts, write fallback dream diaries, use CJK-aware dreaming dedupe, and avoid per-file watcher FD fan-out. ([#80613](https://github.com/openclaw/openclaw/issues/80613), [#82928](https://github.com/openclaw/openclaw/issues/82928), [#85060](https://github.com/openclaw/openclaw/issues/85060), [#85704](https://github.com/openclaw/openclaw/issues/85704), [#85967](https://github.com/openclaw/openclaw/issues/85967), [#86701](https://github.com/openclaw/openclaw/issues/86701)) Thanks [@brokemac79](https://github.com/brokemac79), [@openperf](https://github.com/openperf), and [@yaaboo-gif](https://github.com/yaaboo-gif). - Agents/sessions: include visibility metadata on restricted `sessions_list` results so scoped counts are clearly reported without widening access or exposing hidden-session counts. ([#86944](https://github.com/openclaw/openclaw/issues/86944)) Thanks [@ferminquant](https://github.com/ferminquant). - Gateway/DNS: validate wide-area discovery domains before deriving zone paths or writing zone files, so invalid `discovery.wideArea.domain` and `dns setup --domain` values fail with a DNS-name diagnostic instead of falling through to unrelated configuration errors. Thanks [@mmaps](https://github.com/mmaps). - Agents/BTW: route fallback side-question streams through the embedded stream resolver so Anthropic-compatible MiniMax requests use the same capped transport as normal chat. ([#86312](https://github.com/openclaw/openclaw/issues/86312)) Thanks [@neeravmakwana](https://github.com/neeravmakwana). - Telegram: treat `/command@TargetBot` bot-command entities as explicit mentions for the addressed bot so `requireMention` groups no longer drop targeted commands or captions. Fixes [#84462](https://github.com/openclaw/openclaw/issues/84462). ([#86553](https://github.com/openclaw/openclaw/issues/86553)) Thanks [@luoyanglang](https://github.com/luoyanglang). - CI: bound Docker/Bash E2E tarball npm installs with `OPENCLAW_E2E_NPM_INSTALL_TIMEOUT` so package, onboarding, plugin, and upgrade lanes fail instead of hanging on a stuck npm install. - CI: fail Parallels npm-update smoke jobs after the guest command timeout and cleanup backstop instead of only logging a timeout line. - CI: bound kitchen-sink RPC HTTP probes so stalled gateway readiness or response bodies fail and retry instead of wedging the walker. - CI: bound Telegram user Crabbox proof Bot API calls so stalled Telegram responses fail instead of wedging credential and desktop proof cleanup. - CI: bound MCP channel stdio client initialization so Docker channel proof fails and closes the bridge transport instead of waiting for the outer job timeout. - CI: keep `OPENCLAW_TESTBOX=1 pnpm check:changed` delegating to Blacksmith Testbox through Crabbox without forwarding local Testbox or worker env into the remote command. - CI: send KILL after the TERM grace period for manual checkout fetch timeouts so stuck Testbox and workflow checkout retries cannot hang behind a wedged `git fetch`. - CI: send KILL after the TERM grace period for Bun global install smoke command timeouts so trapped `openclaw` child processes cannot wedge the scheduled install smoke. - iMessage: thread current channel/account inbound attachment roots into the image tool so iMessage-saved attachments under `~/Library/Messages/Attachments` (including the wildcard `/Users/*/Library/Messages/Attachments` root) are read through the existing inbound path policy instead of being rejected as `path-not-allowed`. Literal `localRoots` stays workspace-scoped. Fixes [#30170](https://github.com/openclaw/openclaw/issues/30170). ([#86569](https://github.com/openclaw/openclaw/issues/86569)) - QQ Bot: respect `OPENCLAW_HOME` for outbound media path resolution so `<qqmedia>` sends no longer silently fail when `HOME` and `OPENCLAW_HOME` differ (Docker / multi-user hosts). Persisted QQ Bot data (sessions, known users, refs) stays anchored on the OS home for upgrade compatibility. Fixes [#83562](https://github.com/openclaw/openclaw/issues/83562). Thanks [@sliverp](https://github.com/sliverp). - Update: report the primary malformed `openclaw.extensions` payload error without adding a duplicate missing-main diagnostic. ([#86596](https://github.com/openclaw/openclaw/issues/86596)) Thanks [@ferminquant](https://github.com/ferminquant). - Control UI: keep host-local Markdown file paths inert while preserving app-relative links. ([#86620](https://github.com/openclaw/openclaw/issues/86620)) Thanks [@BryanTegomoh](https://github.com/BryanTegomoh). - Gateway: dampen repeated unauthenticated device-required probes per URL while preserving explicit-auth and paired recovery paths. ([#86575](https://github.com/openclaw/openclaw/issues/86575)) Thanks [@ferminquant](https://github.com/ferminquant). - IRC: store inbound channel routes with the canonical `channel:#name` target and join transient channel sends before writing. ([#85906](https://github.com/openclaw/openclaw/issues/85906)) Thanks [@Kailigithub](https://github.com/Kailigithub). - Usage: surface unknown all-zero model pricing as missing cost entries instead of a confident `$0` total. ([#85882](https://github.com/openclaw/openclaw/issues/85882)) Thanks [@MichaelZelbel](https://github.com/MichaelZelbel). - Agents/Codex: honor yolo app-server approval policy only for the full `never` plus `danger-full-access` case. ([#85909](https://github.com/openclaw/openclaw/issues/85909)) Thanks [@earlvanze](https://github.com/earlvanze). - Gateway/Gmail: clear Gmail watcher renewal intervals on re-entry so hot reloads do not leak lifecycle timers. ([#82947](https://github.com/openclaw/openclaw/issues/82947)) Thanks [@SebTardif](https://github.com/SebTardif). - Logging: exit cleanly on broken stdout/stderr pipes without masking existing failure exit codes. ([#80059](https://github.com/openclaw/openclaw/issues/80059)) Thanks [@pavelzak](https://github.com/pavelzak). - Gateway/security: escape transcript metadata field names while extracting oversized session line prefixes. ([#85934](https://github.com/openclaw/openclaw/issues/85934)) Thanks [@SebTardif](https://github.com/SebTardif). - Plugins/security: validate manifest model pattern regexes with the safe-regex compiler so unsafe patterns are ignored before matching. ([#86046](https://github.com/openclaw/openclaw/issues/86046)) Thanks [@SebTardif](https://github.com/SebTardif). - Discord: route gateway metadata REST lookups through the configured Discord proxy so proxied accounts do not fall back to direct `discord.com` connections before opening the WebSocket. Fixes [#80227](https://github.com/openclaw/openclaw/issues/80227). Thanks [@Clivilwalker](https://github.com/Clivilwalker). - Agents/media: hydrate current-turn image attachments from filename-derived MIME types so active vision can see generated or forwarded images whose source omitted an image content type. ([#84812](https://github.com/openclaw/openclaw/issues/84812)) Thanks [@marchpure](https://github.com/marchpure). - Agents/fs: point workspace-only scratch-path guidance at in-workspace temp directories while keeping host-root writes rejected by the tool guard. ([#86501](https://github.com/openclaw/openclaw/issues/86501)) Thanks [@tianxiaochannel-oss88](https://github.com/tianxiaochannel-oss88). - Agents/media: keep async cron media completions scoped to their run session while preserving direct delivery for stale generated-media success and failure notifications. ([#86529](https://github.com/openclaw/openclaw/issues/86529)) Thanks [@ai-hpc](https://github.com/ai-hpc). - Gateway: emit plugin `session_end`/`session_start` hooks when `agent.send` rotates or replaces a session id, keeping hook lifecycle state aligned with `sessions.changed` notifications. Fixes [#83507](https://github.com/openclaw/openclaw/issues/83507). ([#85875](https://github.com/openclaw/openclaw/issues/85875)) Thanks [@brokemac79](https://github.com/brokemac79). - OpenShell/SSH: reject malformed generated exec commands before sandbox/session setup so unresolved workflow placeholders fail fast instead of reaching the remote shell. Fixes [#72373](https://github.com/openclaw/openclaw/issues/72373). Thanks [@brokemac79](https://github.com/brokemac79). - Google: stop normalizing `gemini-3.1-flash-lite` to the retired preview endpoint and update Flash Lite alias guidance to the GA model id. Fixes [#86151](https://github.com/openclaw/openclaw/issues/86151). ([#86240](https://github.com/openclaw/openclaw/issues/86240)) Thanks [@SebTardif](https://github.com/SebTardif). - Installer: make Alpine apk installs cover Git, verify the Node runtime floor, try `nodejs-current`, and report Alpine version guidance when repositories only provide older Node packages. - Agents/status: prefer the active Claude CLI OAuth auth label over an unused Anthropic env API-key label for equivalent runtime aliases. Fixes [#80184](https://github.com/openclaw/openclaw/issues/80184). ([#86570](https://github.com/openclaw/openclaw/issues/86570)) Thanks [@brokemac79](https://github.com/brokemac79). - Agents/media: send direct fallback for generated media still missing after an active requester wake fails. ([#85489](https://github.com/openclaw/openclaw/issues/85489)) Thanks [@fuller-stack-dev](https://github.com/fuller-stack-dev). - Agents: derive overflow compaction budgets from provider-reported and synthetic over-budget token counts so confirmed context overflows compact before retrying. ([#70473](https://github.com/openclaw/openclaw/issues/70473)) Thanks [@fuller-stack-dev](https://github.com/fuller-stack-dev). - Agents/Codex: recover Codex context-window prompt errors through overflow compaction and surface reset guidance when recovery is exhausted. ([#85542](https://github.com/openclaw/openclaw/issues/85542)) Thanks [@fuller-stack-dev](https://github.com/fuller-stack-dev). - Agents/Codex: allow Codex app-server runs to bootstrap from `CODEX_API_KEY` or `OPENAI_API_KEY` when no Codex auth profile is configured. - Agents/Codex: keep selected Codex runtime routing on OpenAI-Codex while preserving direct OpenAI API-key compaction fallback. ([#86408](https://github.com/openclaw/openclaw/issues/86408)) Thanks [@funmerlin](https://github.com/funmerlin) and [@VACInc](https://github.com/VACInc). - Agent transcript: include OpenClaw agent session logs when finding local transcript candidates. - Crabbox: bootstrap raw AWS macOS shell commands wrapped in absolute `time` paths so RSS probes can run Node and pnpm on fresh macOS runners. - Crabbox: bootstrap raw AWS macOS shell commands even when setup statements precede Node or pnpm usage. - TUI/local: skip unnecessary secret resolution, gateway model catalog loading, bootstrap, and skill scans in explicit local-model runs so startup reaches the model request faster. - Sessions/doctor: load large session stores without clone amplification during read-only doctor checks and reclaim stale `sessions.json.*.tmp` sidecars. Fixes [#56827](https://github.com/openclaw/openclaw/issues/56827). Thanks [@openperf](https://github.com/openperf). - Tests: clean successful plugin gateway gauntlet isolated temp roots while keeping an explicit preservation switch for failed/debug runs. - Plugins/perf: reuse derived plugin metadata snapshots for the lifetime of the process so reply-time skill setup no longer rescans plugin metadata on every turn. - Discord/OpenAI voice: keep wake-name master consults using the current speaker context after ignored ambient transcripts and shorten the default capture silence grace. - Doctor: skip redundant Gateway restart prompts when a recent supervisor restart leaves the Gateway healthy. Fixes [#86518](https://github.com/openclaw/openclaw/issues/86518). ([#86533](https://github.com/openclaw/openclaw/issues/86533)) Thanks [@liaoyl830](https://github.com/liaoyl830). - Cron: restore suspended cron lanes to the configured/default concurrency instead of falling back to one after quota or circuit-breaker auto-resume. - Gateway: keep session-only Control UI tool-start mirrors flowing during diagnostic queue pressure instead of silently dropping non-terminal tool updates. - Agents/memory: return optional not-found context for missing date-only daily memory reads instead of logging benign first-run `ENOENT` failures. Fixes [#82928](https://github.com/openclaw/openclaw/issues/82928). Thanks [@galiniliev](https://github.com/galiniliev). - Discord: merge streamed text captions into following media block replies so captions and attachments send as one message. ([#86487](https://github.com/openclaw/openclaw/issues/86487)) Thanks [@neeravmakwana](https://github.com/neeravmakwana). - Gateway: avoid sending duplicate tool-event frames to Control UI connections that are subscribed by both run and session. - Discord/OpenAI voice: accept broader edge-position fuzzy wake-name transcripts while keeping ambient speech gated. - Discord/OpenAI voice: accept longer leading wake-name mistranscripts such as "Open Club" for OpenClaw. - Agents/OpenAI-compatible: stop ModelStudio-compatible chat requests before sending system/tool-only payloads that have no usable user or assistant turn. ([#86177](https://github.com/openclaw/openclaw/issues/86177)) Thanks [@TurboTheTurtle](https://github.com/TurboTheTurtle). - Gateway/plugins: reuse plugin package realpath checks while building installed plugin indexes so startup avoids repeated filesystem resolution work. - Kilo Gateway: send string `stop` sequences as arrays so Kilo accepts OpenAI-compatible chat completions. ([#86461](https://github.com/openclaw/openclaw/issues/86461)) Thanks [@SebTardif](https://github.com/SebTardif). - Discord/OpenAI voice: accept leading fuzzy wake-name transcripts such as "Monty" or "Moti" for a Molty agent while keeping ambient speech gated. - Media understanding: convert HEIC and HEIF images to JPEG before image description providers run so iPhone photos work in direct and configured image-description flows. ([#86037](https://github.com/openclaw/openclaw/issues/86037)) - Agents: release embedded-attempt session locks from outer teardown so post-prompt exceptions cannot wedge later requests behind `SessionWriteLockTimeoutError`. Fixes [#86014](https://github.com/openclaw/openclaw/issues/86014). Thanks [@openperf](https://github.com/openperf). - Discord/OpenAI voice: rotate Realtime sessions at provider max duration without logging the expected session-expiry event as an error. - Sessions: skip metadata-only entries during QMD-slugified session lookup so one incomplete row does not block transcript hit resolution. ([#86327](https://github.com/openclaw/openclaw/issues/86327)) Thanks [@abnershang](https://github.com/abnershang). - Agents/media: derive bundled plugin local-media trust from plugin tool metadata instead of importing the full plugin registry on subscription paths. ([#84409](https://github.com/openclaw/openclaw/issues/84409)) Thanks [@samzong](https://github.com/samzong). - Image tool: keep config-backed custom-provider API keys usable for auto-discovered vision models, including deferred image-tool execution without env keys or auth profiles. ([#85733](https://github.com/openclaw/openclaw/issues/85733)) - Memory/local embeddings: run local GGUF embeddings in an isolated worker sidecar and degrade to configured fallback or keyword search on worker failure so native embedding crashes do not take down the Gateway. ([#85348](https://github.com/openclaw/openclaw/issues/85348)) Thanks [@osolmaz](https://github.com/osolmaz). - Gateway: clear the runtime config snapshot before `SIGUSR1` in-process restarts so config changes survive the next gateway loop. ([#86388](https://github.com/openclaw/openclaw/issues/86388)) Thanks [@XuZehan-iCenter](https://github.com/XuZehan-iCenter). - Models: show OAuth delegation markers as configured `models.json` auth while keeping runtime route usability checks strict. ([#86378](https://github.com/openclaw/openclaw/issues/86378)) Thanks [@rohitjavvadi](https://github.com/rohitjavvadi). - Cron: seed active scheduled and manual cron task rows with a progress summary so status surfaces do not look blank while jobs run. ([#86313](https://github.com/openclaw/openclaw/issues/86313)) Thanks [@ferminquant](https://github.com/ferminquant). - Cron: preserve unsupported persisted cron payload rows during routine store writes while keeping those rows non-runnable. Fixes [#84922](https://github.com/openclaw/openclaw/issues/84922). ([#86415](https://github.com/openclaw/openclaw/issues/86415)) Thanks [@IWhatsskill](https://github.com/IWhatsskill). - Updater: exclude prerelease git tags from stable channel resolution so source updates do not check out newer alpha/rc/preview/canary tags. ([#86260](https://github.com/openclaw/openclaw/issues/86260)) Thanks [@stevenepalmer](https://github.com/stevenepalmer). - Security/Audit: flag webhook `hooks.token` reuse of active Gateway password auth in `openclaw security audit` while keeping password-mode startup compatibility. ([#84338](https://github.com/openclaw/openclaw/issues/84338)) Thanks [@coygeek](https://github.com/coygeek). - QQBot: derive the outbound reply watchdog from configured agent and provider timeouts so slow local model replies are not cut off at five minutes. Fixes [#85267](https://github.com/openclaw/openclaw/issues/85267). ([#85271](https://github.com/openclaw/openclaw/issues/85271)) Thanks [@SymbolStar](https://github.com/SymbolStar). - Agents/heartbeat: stop heartbeat turns after the first valid `heartbeat_respond` so repeated response loops do not burn tokens. ([#86357](https://github.com/openclaw/openclaw/issues/86357)) Thanks [@udaymanish6](https://github.com/udaymanish6). - Tasks: keep retained lost tasks out of default status health counts, explain their cleanup window during maintenance, and prune lost task records after 24 hours instead of the general 7-day terminal retention. - Memory-core: keep REM dreaming focused on live light-staged memories and mark staged entries as considered so old recall history no longer dominates fresh candidates. ([#86302](https://github.com/openclaw/openclaw/issues/86302)) Thanks [@SebTardif](https://github.com/SebTardif). - Memory: abort sync instead of downgrading an existing semantic vector index to FTS-only when the configured embedding provider is temporarily unavailable. ([#85704](https://github.com/openclaw/openclaw/issues/85704)) Thanks [@yaaboo-gif](https://github.com/yaaboo-gif). - Telegram: propagate forum topic names through the account-scoped topic cache for native command context and topic create/edit actions. ([#86299](https://github.com/openclaw/openclaw/issues/86299)) Thanks [@SebTardif](https://github.com/SebTardif). - Slack: keep downloaded read-only files out of reply media so Slack file reads do not echo files back to the conversation. ([#86318](https://github.com/openclaw/openclaw/issues/86318)) Thanks [@neeravmakwana](https://github.com/neeravmakwana). - Cron: accept leading-plus relative durations such as `+5m` for one-shot `--at` schedules. ([#86341](https://github.com/openclaw/openclaw/issues/86341)) Thanks [@mushuiyu886](https://github.com/mushuiyu886). - Agents/media: preserve async-started media tool metadata so background generation starts no longer surface generic incomplete-turn warnings while replay stays unsafe. ([#85933](https://github.com/openclaw/openclaw/issues/85933)) Thanks [@fuller-stack-dev](https://github.com/fuller-stack-dev). - Docker E2E: dedupe scheduler lane resources so npm/service package lanes are not over-counted and serialized unnecessarily. - QA/diagnostics: add a collector-backed OpenTelemetry smoke lane, make the OTLP payload leak check scenario-aware, and keep source QA builds from failing on optional dependency imports resolved through pnpm's temp module path. - Crabbox: bootstrap Git metadata for sparse remote changed gates so raw synced workspaces can run `pnpm check:changed` from the intended diff. - xAI/LM Studio: avoid buffering ordinary bracketed or `final` prose until stream completion while watching for plain-text tool-call fallbacks. - Doctor: warn and continue when the cron job store exists but cannot be read so later health checks still run. Fixes [#86102](https://github.com/openclaw/openclaw/issues/86102). ([#86384](https://github.com/openclaw/openclaw/issues/86384)) Thanks [@1052326311](https://github.com/1052326311). - Discord: suppress a bot's previous reply body and referenced media from prompt context when a user replies to that bot message, while keeping reply metadata for routing. ([#86238](https://github.com/openclaw/openclaw/issues/86238)) Thanks [@fuller-stack-dev](https://github.com/fuller-stack-dev). - Discord: restore bare numeric channel IDs for outbound message-tool sends while keeping explicit DM targets unambiguous. ([#86571](https://github.com/openclaw/openclaw/issues/86571)) Thanks [@joshavant](https://github.com/joshavant). - Docker E2E: avoid rebuilding the Control UI twice while preparing the shared OpenClaw package tarball for package-backed scenario runs. - Tests: avoid rebuilding the Control UI twice during the installer Docker smoke now that `pnpm build` includes `ui:build`. - Tests: give QA config mutation RPCs enough native Windows budget to finish gateway config writes and restart settle after hot scenario runs. - Tests: keep the gateway restart-inflight QA scenario focused on restart recovery on native Windows by allowing expected embedded prompt handoff errors and using the Windows-safe timeout budget. - QA-Lab: make the synthetic OpenAI provider honor generic `reply exactly:` directives after required kickoff reads so restart-recovery scenarios do not fall through to generic repo-summary prose. - Gateway: abort active `agent` RPC runs during forced restart shutdown so stale in-process turns cannot keep writing a session after the Gateway lifecycle restarts. - Crabbox: sync clean sparse worktrees through a temporary full checkout even when reusing an existing lease so tracked build-time files are not omitted. - Build: route `scripts/ui.js` through the shared pnpm runner and keep Control UI chunking helpers in sparse-included source so native Windows Corepack builds can produce `dist/control-ui`. - Tests: give the memory fallback QA scenario enough turn budget to exercise native Windows gateway runs instead of failing on the client timeout while the mock agent is still dispatching. - Tests: collect QA gateway CPU/RSS metrics on native Windows and give the channel baseline enough turn budget to report slow gateway runs instead of timing out before proof. - Install/update: bypass npm `min-release-age` policies with `--min-release-age=0` instead of `--before` so hosted installers keep working on npm versions that reject the combined config. ([#84749](https://github.com/openclaw/openclaw/issues/84749)) Thanks [@TeodoroRodrigo](https://github.com/TeodoroRodrigo). - Diagnostics: reclaim wedged session lanes when stale active-run bookkeeping blocks queued work despite no forward progress. Fixes [#85639](https://github.com/openclaw/openclaw/issues/85639). Thanks [@openperf](https://github.com/openperf). - WebChat: keep message-tool replies visible in the chat while still summarizing internal tool results for the model. Fixes [#86347](https://github.com/openclaw/openclaw/issues/86347). Thanks [@shakkernerd](https://github.com/shakkernerd). - Gateway/perf: fail startup benchmark samples when the Gateway process exits before benchmark teardown, including signal deaths after readiness probes. - Gateway/perf: fail restart benchmark samples when the Gateway exits before benchmark teardown, including clean exits and signal deaths after successful restart probes. - Agents/tests: keep model catalog visibility on static selection helpers so catalog visibility checks avoid the broad model-selection barrel import. - Agents/commitments: serialize commitment store load-modify-save writes so concurrent heartbeat and CLI updates no longer lose dismissal, sent, or attempt state. ([#81153](https://github.com/openclaw/openclaw/issues/81153)) Thanks [@ai-hpc](https://github.com/ai-hpc). - xAI/LM Studio: promote plain-text tool-call fallbacks into structured tool calls and strip leaked internal tool syntax before user-facing delivery. ([#86222](https://github.com/openclaw/openclaw/issues/86222)) Thanks [@fuller-stack-dev](https://github.com/fuller-stack-dev). - CLI: suppress benign self-update version-skew warnings during package post-update finalization. - Gateway/perf: tighten restart and startup benchmark failure handling so long profiling runs, failed probes, and fresh Linux runners no longer produce false passing or `n/a` results. - Checks: keep intentional Knip unused-file findings optional so full CI and sparse proof workspaces stay aligned. - Docker: restore writable `~/.config` in runtime images. Fixes [#85968](https://github.com/openclaw/openclaw/issues/85968). Thanks [@hkoessler](https://github.com/hkoessler) and [@Bartok9](https://github.com/Bartok9). - Plugin SDK: keep legacy root diagnostic subscriptions connected when built plugin SDK aliases resolve diagnostic helpers through a separate module graph. - Diagnostics: export alertable OTel and Prometheus signals for blocked tools, model failover, stale sessions, liveness warnings, oversized payloads, and webhook ingress while fixing shared OTLP endpoints with query strings. - Tests: normalize macOS canonical temp paths in exec allowlists, fs-safe trash assertions, installed plugin matching, Telegram topic-name stores, and built ACPX MCP server expectations so native macOS proof runners cover the intended behavior. - Codex/app-server: preserve message-tool-only source reply delivery mode on active runs so sub-agent completion wakeups can steer the active Codex turn instead of being rejected. ([#86287](https://github.com/openclaw/openclaw/issues/86287)) Thanks [@ferminquant](https://github.com/ferminquant). - Tests: sample the Windows kitchen-sink RPC gateway directly and serialize RSS probes so native runs keep the memory guard active. - Tests: normalize bundled plugin lifecycle probe paths and state-root lookup so native Windows release sweeps accept valid packaged plugin installs. - Agents/Claude CLI: route live native Bash permission requests through OpenClaw exec policy so Claude turns no longer stall on `control_request`, and document that OpenClaw exec policy is authoritative. Fixes [#80819](https://github.com/openclaw/openclaw/issues/80819). ([#86330](https://github.com/openclaw/openclaw/issues/86330), from [#81971](https://github.com/openclaw/openclaw/issues/81971)) Thanks [@guthirry](https://github.com/guthirry) and [@sallyom](https://github.com/sallyom). - Security audit: warn when YOLO OpenClaw exec policy overrides a restrictive raw Claude `--permission-mode` for managed live sessions. ([#86557](https://github.com/openclaw/openclaw/issues/86557)) Thanks [@sallyom](https://github.com/sallyom). - Config: keep benign legacy metadata write anomalies out of default doctor and config command output while preserving explicit anomaly logging for diagnostics. - Codex: log when implicit app-server `never` approvals are promoted for OpenClaw tool policy, including whether the trigger was a `before_tool_call` hook or trusted tool policy. - Codex harness: make subscription usage-limit errors without reset times explain that OpenClaw cannot determine the reset and point users to wait until Codex is available, use another Codex account, or switch to another configured model/provider. Thanks [@amknight](https://github.com/amknight). - Google Vertex: support production ADC modes such as Workload Identity Federation, service-account credentials, and metadata-server ADC for the native Vertex transport. ([#83971](https://github.com/openclaw/openclaw/issues/83971)) Thanks [@damianFelixPago](https://github.com/damianFelixPago). - Telegram: route normal `[telegram][diag]` polling diagnostics through `runtime.log` while keeping non-diag warnings and persistence failures on `runtime.error`, so healthy polling startup no longer looks like an error. Fixes [#82957](https://github.com/openclaw/openclaw/issues/82957). ([#82958](https://github.com/openclaw/openclaw/issues/82958)) Thanks [@galiniliev](https://github.com/galiniliev). - Providers/Ollama: strip inline Kimi cloud reasoning prefixes from streamed and final visible replies while keeping ordinary Kimi answers append-only. ([#86286](https://github.com/openclaw/openclaw/issues/86286)) Thanks [@jason-allen-oneal](https://github.com/jason-allen-oneal). - Gateway: require Talk secret authority before setup-code handoff can include Talk secrets. ([#85690](https://github.com/openclaw/openclaw/issues/85690)) Thanks [@ngutman](https://github.com/ngutman). - Agents: keep fallback error reporting scoped to the active model candidate so stale prior-provider quota/auth text is not reported for later fallback attempts. ([#86134](https://github.com/openclaw/openclaw/issues/86134)) Thanks [@zhangguiping-xydt](https://github.com/zhangguiping-xydt). - iMessage: dedupe watcher startup when `channels.imessage.accounts` lists both `default` and a named account that point at the same local Messages source, so the gateway no longer spawns two `imsg rpc` processes or doubles inbound replies; the dedupe is scoped to watcher startup, leaving duplicate accounts addressable for outbound sends, status, and capability listings, and `openclaw doctor` flags the redundant account with a rebinding hint. Fixes [#65141](https://github.com/openclaw/openclaw/issues/65141). ([#86705](https://github.com/openclaw/openclaw/issues/86705)) Thanks [@swang430](https://github.com/swang430). </details> --- ### Configuration 📅 **Schedule**: Branch creation - At any time (no schedule defined), Automerge - At any time (no schedule defined). 🚦 **Automerge**: Disabled by config. Please merge this manually once you are satisfied. ♻ **Rebasing**: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox. 🔕 **Ignore**: Close this PR and you won't be reminded about these updates again. --- - [ ] If you want to rebase/retry this PR, check this box --- This PR has been generated by [Renovate Bot](https://github.com/renovatebot/renovate).  Reviewed-on: https://git.erwanleboucher.dev/eleboucher/homelab/pulls/682

* fix(memory): reject prompt-like memory stores * fix(changelog): mention memory store rejection

openclaw-barnacle Bot added extensions: memory-lancedb Extension: memory-lancedb size: S maintainer Maintainer-authored PR labels May 27, 2026

eleqtrizit self-assigned this May 27, 2026

fix(memory): reject prompt-like memory stores

304024b

eleqtrizit force-pushed the 692 branch from b1cebd7 to 304024b Compare May 27, 2026 04:12

fix(changelog): mention memory store rejection

7e82e49

eleqtrizit merged commit 119d235 into openclaw:main May 27, 2026
89 of 90 checks passed

github-actions Bot pushed a commit to Desicool/openclaw that referenced this pull request May 27, 2026

fix(memory): reject prompt-like memory stores (openclaw#87142)

7aa8283

* fix(memory): reject prompt-like memory stores * fix(changelog): mention memory store rejection

eleqtrizit deleted the 692 branch May 28, 2026 20:20

SYU8384 pushed a commit to SYU8384/openclaw that referenced this pull request Jun 3, 2026

fix(memory): reject prompt-like memory stores (openclaw#87142)

e3332ac

* fix(memory): reject prompt-like memory stores * fix(changelog): mention memory store rejection

sablehead pushed a commit to sablehead/openclaw that referenced this pull request Jun 10, 2026

fix(memory): reject prompt-like memory stores (openclaw#87142)

d87d50c

* fix(memory): reject prompt-like memory stores * fix(changelog): mention memory store rejection

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

fix(memory): reject prompt-like memory stores#87142

fix(memory): reject prompt-like memory stores#87142
eleqtrizit merged 2 commits into
openclaw:mainfrom
eleqtrizit:692

eleqtrizit commented May 27, 2026 •

edited

Loading

Uh oh!

clawsweeper Bot commented May 27, 2026 •

edited

Loading

Uh oh!

clawsweeper Bot commented May 27, 2026 •

edited

Loading

Uh oh!

eleqtrizit commented May 27, 2026

Uh oh!

clawsweeper Bot commented May 27, 2026 •

edited

Loading

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

eleqtrizit commented May 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Changes

Validation

Notes

Real behavior proof

Uh oh!

clawsweeper Bot commented May 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

clawsweeper Bot commented May 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Hatch command

Uh oh!

eleqtrizit commented May 27, 2026

Uh oh!

clawsweeper Bot commented May 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

eleqtrizit commented May 27, 2026 •

edited

Loading

clawsweeper Bot commented May 27, 2026 •

edited

Loading

clawsweeper Bot commented May 27, 2026 •

edited

Loading

clawsweeper Bot commented May 27, 2026 •

edited

Loading