* fix(hooks): flag hook event names that no trigger site emits
* docs(hooks): clarify bare family subscriptions in unknown-event note
* fix(hooks): word unknown-event diagnostics around core-emitted keys
* fix(hooks): apply unknown-event advisory to legacy config handlers too
* improve(health): surface dead-lettered delivery queue entries
Deliveries that exhaust retries are moved to the failed status in the
SQLite delivery queue for diagnostics, but no health surface ever read
them back: openclaw health stayed all-green while messages sat
dead-lettered, visible only in gateway logs. Add a per-queue failed
count accessor and report dead-lettered entries in the health snapshot
(JSON field plus a warning line), following the existing plugin and
context-engine health section pattern. Observer-only: no retry, purge,
or behavior change.
* test(health): cover snapshot and CLI dead-letter reporting
Add getHealthSnapshot coverage against an isolated state dir, an
openclaw health text-output test for the warning line, exact
oldest-failure timestamp assertions via fake timers, and a debug log
when the delivery-queue health read fails.
* fix(health): recompute dead-letter counts for cached gateway responses
The gateway health handler can serve a cached snapshot for up to a
refresh interval, so a delivery dead-lettered after the cache was
filled stayed hidden from openclaw health. Recompute the delivery
queue summary in mergeCachedHealthRuntimeState alongside the existing
context-engine and model-pricing live merges, and cover the cached
handler path with a regression test.
* fix(usage-bar): bound template file cache to prevent unbounded watcher growth
Add MAX_CACHED_TEMPLATE_FILES=64 limit; evict the oldest entry (closing
its fs.watch watcher) before allocating a watcher for a new key when
the cache is full.
Eviction runs before watcher allocation so we never create a watcher
only to close it immediately. Eviction triggers only when inserting a
new key (!fileCache.has(path)) — retries for an existing key must
not evict other entries.
Fixes#98960
Co-Authored-By: Claude <noreply@anthropic.com>
* refactor(test): replace mutable dir variable with cleanup-stack pattern in template.test.ts
Co-Authored-By: Claude <noreply@anthropic.com>
* chore(test): add curly braces for eslint curly rule
Co-Authored-By: Claude <noreply@anthropic.com>
* test(usage-bar): register watcher proof cleanup
---------
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Vincent Koc <vincentkoc@ieee.org>
* fix(transcripts): handle read stream errors gracefully in TranscriptsStore
* proof(transcripts): add real behavior proof script for store stream error catch
* fix(transcripts): reject with Error in stream error handler
* proof(transcripts): replace wrapper with real EISDIR stream error proof
* fix(transcripts): reject non-ENOENT stream errors after stream close
* fix(feishu): bound Feishu API JSON response reads to prevent OOM
Replace the bare `response.json()` in `fetchFeishuJson`
(`extensions/feishu/src/app-registration.ts`) with a `readResponseWithLimit`
call capped at 16 MiB. A misconfigured or adversarial Feishu endpoint that
streams an unbounded body previously had no defence; the bounded reader now
cancels the stream at the cap and throws a labelled `feishu.api` error.
Tests: over-cap (32 MiB stream, no Content-Length — stream cancelled, error
matches feishu.api), under-cap (chunked valid JSON — parsed and returned),
and malformed-JSON (labelled feishu.api error). All five tests pass.
* fix(feishu): delegate JSON reads to provider helper
* test(feishu): add real node:http server proof for readProviderJsonResponse bound
* test(feishu): fix lint errors in real HTTP server proof (curly + no-promise-executor-return)
* test(feishu): prove bound reads through SSRF guard
* fix(feishu): satisfy overloaded LookupFn type in hermetic test lookup stub
* fix(mistral): correct model id typo in usesReasoningEffort helper
* chore: re-trigger CI
* test(mistral): cover medium reasoning effort
---------
Co-authored-by: Peter Steinberger <steipete@gmail.com>
The markdown link regex character class [^)\s>]+ stopped at the first
closing paren, truncating URLs like wikipedia.org/wiki/URL_(disambiguation)
at the opening (. Replace with a balanced-parens pattern that supports
one level of parenthetical groups inside the URL path.
Add trimUnbalancedTrailingParens for bare URLs in prose where a trailing
) belongs to surrounding punctuation rather than the URL itself.
* fix(tools-manager): replace spawnSync extraction with safe extractArchive API
Replace the synchronous spawnSync-based archive extraction (unzip/tar)
with the safe extractArchive from @openclaw/fs-safe, which enforces:
- maxArchiveBytes: 100 MB (prevents oversized compressed input)
- maxExtractedBytes: 500 MB (prevents decompression bomb OOM)
- maxEntries: 1000 (prevents zip bomb from exhausting inodes)
- timeoutMs: 60,000 (prevents hung extraction)
- Path traversal and symlink/hardlink protection (built-in)
Removes 5 functions (formatSpawnFailure, runExtractionCommand,
extractTarGzArchive, getWindowsTarCommand, extractZipArchive) and their
associated imports, replacing them with a single async wrapper around
the existing @openclaw/fs-safe infrastructure (re-exported from
src/infra/archive.ts).
Net: -83 lines, + security boundaries across all platforms.
* fix(tools-manager): add download-stream byte cap and regression tests
Add a maxBytes parameter to downloadFile that checks Content-Length
before reading the body and enforces a streaming byte cap during
transfer, so oversized archives are rejected before hitting disk.
Extract MAX_ARCHIVE_BYTES as a module-level constant shared between
downloadFile and extractArchiveSafe, ensuring both gates use the same
100 MB limit.
Add two regression tests:
- rejects downloads with Content-Length exceeding the archive byte cap
- accepts downloads with Content-Length under the archive byte cap
Ref. https://github.com/openclaw/openclaw/pull/98988
* fix(tools-manager): replace PassThrough with Transform stream limiter
- Replace PassThrough data listener with Transform that rejects overflow
chunks *before* they are forwarded to the file pipeline, preventing the
offending chunk from landing on disk.
- Add completion guard and cleanup of partial downloads on failure so
downloadTool does not leave a partial archive on disk.
- Add real behavior proof: test output and live fd/rg download and
extraction through the safe extractArchive code path.
🦞 diamond lobster: L2 evidence (real function calls + real objects)
Ref. https://github.com/openclaw/openclaw/pull/98988
* fix(tools-manager): replace PassThrough with Transform stream limiter
Uses a Transform to reject overflow chunks *before* they are forwarded
to the file pipeline (a PassThrough data listener acts *after* emission
and cannot prevent the offending chunk from landing on disk).
Wraps the pipeline in a completion-guarded block so that partial
downloads are removed when the transfer fails mid-way.
Ref. https://github.com/openclaw/openclaw/pull/98988
* fix(agents): harden helper archive downloads
* fix(agents): preserve archive extraction cause
---------
Co-authored-by: Vincent Koc <25068+vincentkoc@users.noreply.github.com>
Co-authored-by: Vincent Koc <vincentkoc@ieee.org>
* fix(infra): include update timeout in managed-service handoff parent exit wait
The handoff parent exit grace was hard-coded to 60s regardless of
the overall update timeout. When the gateway is draining active
tasks, 60s is often not enough, causing
managed-service-handoff-parent-timeout.
Use the existing timeoutMs parameter as a floor for the parent
exit wait so callers can extend it for long-running drains.
Refs #99666
* fix(infra): add shutdown reserve to managed-service handoff parent exit wait
* fix(infra): align update handoff with restart drain
Co-authored-by: 徐闻涵0668001344 <xu.wenhan1@xydigit.com>
* fix(infra): arm restart with managed handoff
* fix(infra): use live restart drain for auto-update
* fix(infra): force process exit for managed updates
* fix(infra): normalize managed handoff restart delay
* test(infra): cover handoff drain argument
* fix(gateway): upgrade accepted restarts for managed updates
* fix(gateway): bound managed update restart upgrades
* test(gateway): stabilize managed update handoff proof
---------
Co-authored-by: Peter Steinberger <steipete@gmail.com>
* fix(agents): prefer current attempt terminal state
Co-authored-by: weiqinl <liu.weiqin@xydigit.com>
* test(agents): model current attempt terminal state
---------
Co-authored-by: weiqinl <liu.weiqin@xydigit.com>
* fix(agents): keep background exec exit notifications UTF-16 safe at the snippet cut
Background exec exit notifications truncate the command's tail output with a
raw String.slice, which can split a UTF-16 surrogate pair when an emoji lands
at the cut. Use the existing truncateUtf16Safe helper so the message delivered
to the channel never ends with a dangling surrogate half.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(agents): protect exec notify tail boundaries
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(agents): protect node exec follow-up tails
Co-authored-by: Cursor <cursoragent@cursor.com>
* test(agents): type node exec follow-up assertions
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(agents): keep bounded exec output Unicode-safe
* docs(changelog): note exec output Unicode safety
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Peter Steinberger <steipete@gmail.com>
* fix(memory-core): guard qmd mcporter JSON.parse against non-JSON stdout
runQmdSearchViaMcporter parsed mcporter subprocess stdout with JSON.parse
outside the runMcporter try/catch (qmd-manager.ts:2722). A non-JSON stdout
(daemon warning bleeding onto stdout, output truncated by maxOutputChars, CLI
killed early, or flag mismatch) threw a raw SyntaxError that propagated
uncaught out of runQmdSearchViaMcporter, surfacing in agent logs as a
context-free SyntaxError with no hint of the actual mcporter failure.
Wrap JSON.parse in try/catch and throw a typed Error carrying a stdout snippet
(matching the guard pattern already used in parseListedCollections in this
file, and the recent matrix #97973 / sms #97999 / signal #98073 /
telegram-ingress #98372 JSON.parse guard series).
Co-Authored-By: Claude <noreply@anthropic.com>
* fix(memory-core): preserve cause in qmd mcporter JSON.parse guard
Add { cause: err } to the re-thrown Error to satisfy the preserve-caught-error
lint rule; the original SyntaxError is now chained, improving diagnosability.
Co-Authored-By: Claude <noreply@anthropic.com>
* fix(memory-core): redact raw stdout from qmd mcporter error (security-boundary)
ClawSweeper flagged that the previous error message exposed raw mcporter
stdout (first 200 chars) before session visibility filtering, which could
leak sensitive content. Drop the stdout preview from the thrown message;
keep the original SyntaxError as `cause` for diagnostics so the parse-failure
reason is still reachable without surfacing unfiltered subprocess output.
Co-Authored-By: Claude <noreply@anthropic.com>
* fix(memory-core): keep qmd mcporter error message generic (no raw stdout leak)
The SyntaxError thrown by JSON.parse embeds a snippet of the raw input in its
message (e.g. Unexpected token '<', then the raw bytes). Including that
SyntaxError message in the thrown Error would surface unfiltered mcporter
stdout before session visibility filtering. Drop the parse-error message from
the thrown Error; keep the original SyntaxError as cause for developer
diagnostics only.
Co-Authored-By: Claude <noreply@anthropic.com>
* fix(memory-core): keep qmd mcporter cause generic (no raw stdout leak via formatErrorMessage)
formatErrorMessage walks the .cause chain into the user-visible path.
Keeping the JSON.parse SyntaxError on .cause leaked its embedded raw
stdout snippet through formatErrorMessage even with a generic message.
Give the cause a generic message too; the raw snippet no longer reaches
the user.
Co-Authored-By: Claude <noreply@anthropic.com>
---------
Co-authored-by: Claude <noreply@anthropic.com>
