docs(changelog): format the merged 13094 fragment as a markdown bullet by anhtahaylove · Pull Request #13158 · diegosouzapw/OmniRoute

anhtahaylove · 2026-09-10T02:32:10Z

The changelog fragment added by #13097 shipped as a bare paragraph instead of a markdown bullet.

scripts/check/check-changelog-integrity.mjs requires every fragment to start with - , so this fails the Merge integrity (changelog + generated skills) gate on every branch built off release/v3.8.51 — not only the PR that introduced it.

✗ changelog.d/fixes/13094-deepseek-pow-slot-leak.md: fragment must start with a markdown bullet ("- ")

Reformatted to match the surrounding fragments (bolded scope prefix + PR link). The gate reports 0 violations afterwards. Content is unchanged apart from the bullet and the link.

The fragment shipped as a bare paragraph, so check-changelog-integrity fails on every branch built off release/v3.8.51, not just the PR that added it.

anhtahaylove · 2026-09-10T04:03:18Z

These two are the smallest PRs in the queue and they unblock every other open PR, so they are worth taking first.

Both gates below are red on release/v3.8.51 itself, not caused by any feature branch. Every PR cut from the base inherits the failure:

Gate	Cause on base	Fixed by
Merge integrity (changelog)	`changelog.d/fixes/13094-deepseek-pow-slot-leak.md` is a bare paragraph; the checker requires a `-` bullet	this PR (#13158)
Docs Gates	`README.md` / `AGENTS.md` / `llm.txt` say "171 migrations"; `src/lib/db/migrations` holds 172 `.sql` files	#13160

Verified by checking out a clean release/v3.8.51 (detached) and running the same commands CI runs — both reproduce with no patch applied. With both PRs applied on top of that clean base:

npm run check:changelog-integrity -> 0 violations
npm run check:docs-counts -> exit 0, no STRICT drift (2 pre-existing soft warnings remain, non-blocking)

Neither is sufficient alone: each clears one gate while still inheriting the other from the base, so #13158 and #13160 should land together.

Note on #13094: that fragment came from my own merged PR #13097. It passed at the time because Quality Gates runs with continue-on-error: true for fork PRs, so the malformed fragment was never blocking on the PR that introduced it — it only became a base-red after merge. Apologies for the extra round trip.

No code changes in either PR: one markdown fragment reformatted, three docs counts corrected.

anhtahaylove · 2026-09-10T07:01:12Z

Review queue: 14 open PRs from a resource-leak audit

All 14 are mergeable and independent of each other, but two of them unblock the other twelve, so the order matters more than the volume.

Please start here — these two are one-line docs fixes that unblock everything else

PR	Fix	Why it blocks others
#13158	Reformat a merged changelog fragment as a markdown bullet	`Merge integrity` gate fails on the merge result for every PR
#13160	Correct migration count 171 → 172	`Docs Gates` fails on every PR

Both failures come from release/v3.8.51 itself, not from the PRs sitting under them. Verified by cherry-picking the two onto a clean base: check:changelog-integrity → 0 violations, check:docs-counts → exit 0. Until they land, all twelve below show red gates that have nothing to do with their content.

The twelve leak fixes

Each is one leak, one file, with a regression test that fails without the fix:

PR	Leak
#13091	Compression pool: idle workers never terminated on eviction
#13092	Plugins: an exit listener leaked per SIGKILL escalation
#13093	LLMLingua: worker spawn silently disabled compression
#13096	ACP: listeners/timers/sessions retained on every `sendPrompt` outcome
#13100	ACP: unbounded session output buffers
#13106	Gamification: badge SSE stream left open on an aborted signal
#13109	DB: process listeners retained when a `node:sqlite` adapter closes
#13114	CLI helper: log stream timeout never cleared on abort
#13155	Traffic inspector: WS subscriber + ping timer survive a dead socket
#13166	Telegram: unbounded per-user API key cache
#13171	Stream: upstream body never released when the JSON→SSE sniff times out
#13175	Security — Telegram webhook accepted unauthenticated deliveries

#13175 is the one worth prioritising after the two base fixes. The webhook path took chat.id straight from an unauthenticated request body, and that id reaches proxyChat(), which mints a real API key and spends upstream quota. It is a breaking change for existing webhook deployments (they must set TELEGRAM_WEBHOOK_SECRET and re-run setWebhook), which is deliberate — the alternative is continuing to serve an open endpoint.

On the red CI you'll see

Beyond the two base failures above, Unit Tests fast-path is also red — but it is flaky on this base, not caused by these PRs. The evidence: #13160 changes one number in three markdown files and cannot affect runtime behaviour, yet its unit-test job fails with a completely different set of tests than #13175's. Two PRs, disjoint failure sets, one of them docs-only. I've kept a per-run record if it's useful for triage.

Also closed #12815, #12820 and #12823 today — superseded by #13091, #13092 and #13093, which carry the identical source change plus tests. Nothing lost, three fewer PRs to look at.

Happy to split, rebase, or re-target any of these if a different shape is easier to review.

anhtahaylove · 2026-09-10T19:52:32Z

Closing as obsolete — the base fixed itself.

release/v3.8.51 has moved 36 commits since this PR was opened, and the fragment it targets has already been reformatted upstream. On current base:

$ git show origin/release/v3.8.51:changelog.d/fixes/13094-deepseek-pow-slot-leak.md
- Fix a concurrency-slot leak in the DeepSeek PoW solver: ...

The leading - bullet — the whole point of this PR — is already there, which is also why the PR now shows as CONFLICTING.

Verified the gate directly on current base:

$ npm run -s check:changelog-integrity
[changelog-integrity] OK - no base bullets lost vs origin/release/v3.8.51.
EXIT=0

So this no longer blocks anything. Same story for #13160 (migration count is already 172 on base). All 15 remaining PRs report MERGEABLE and are no longer gated on these two.

Sorry for the earlier review nudges on both — they were accurate when posted, but the base has since absorbed the fix.

@anhtahaylove

…ady aborted (#13106) Correct: an `abort` listener registered on an already-aborted signal never fires, and `safeEnqueue` can't save it because enqueuing into an unread stream only buffers. The abort-later test earning its keep as a guard on the healthy path is the right instinct. I reformatted the changelog fragment to the `changelog.d` convention (`- **fix(scope):** …`) before merging — `check:changelog-integrity` rejects a fragment that does not start with a markdown bullet, which is the same gate your #13158 was about. Wording is yours, unchanged in substance. --- Validated in one consolidated worktree cut from `release/v3.8.51`, boarded together with the other 13 PRs of this batch — zero merge conflicts between them. - `typecheck:core` clean - complexity 2799 / baseline 3218 and cognitive-complexity 1265 / baseline 1437 — both under baseline - 71 focused assertions green across the 13 test files this batch adds or touches ⚠️ base-red inherited: #12732 — `Docs Gates (fast-path)`, `Merge integrity`, `No new ESLint warnings`, `Unit Tests fast-path` and `Fast Quality Gates` all reproduce on the pure `release/v3.8.51` tip (provider count 356 vs the 358 the modules define, SKILL.md drift, and `open-sse/utils/stream.ts` at 3115 > frozen 3098). None of them touch this diff. Thanks @anhtahaylove — the root-cause write-up, the measured before/after numbers and the red-before-green proof on every one of these made the batch reviewable as a unit.

@anhtahaylove

…13109) Right precedent — #7494 fixed exactly this for the sql.js adapter and the `node:sqlite` one never got the same treatment, even though it is the default driver whenever better-sqlite3 is unavailable. `/api/db-backups/import` opening a throwaway adapter per request makes it reachable. I reformatted the changelog fragment to the `changelog.d` convention (`- **fix(scope):** …`) before merging — `check:changelog-integrity` rejects a fragment that does not start with a markdown bullet, which is the same gate your #13158 was about. Wording is yours, unchanged in substance. --- Validated in one consolidated worktree cut from `release/v3.8.51`, boarded together with the other 13 PRs of this batch — zero merge conflicts between them. - `typecheck:core` clean - complexity 2799 / baseline 3218 and cognitive-complexity 1265 / baseline 1437 — both under baseline - 71 focused assertions green across the 13 test files this batch adds or touches ⚠️ base-red inherited: #12732 — `Docs Gates (fast-path)`, `Merge integrity`, `No new ESLint warnings`, `Unit Tests fast-path` and `Fast Quality Gates` all reproduce on the pure `release/v3.8.51` tip (provider count 356 vs the 358 the modules define, SKILL.md drift, and `open-sse/utils/stream.ts` at 3115 > frozen 3098). None of them touch this diff. Thanks @anhtahaylove — the root-cause write-up, the measured before/after numbers and the red-before-green proof on every one of these made the batch reviewable as a unit.

@anhtahaylove

`stop()` is the normal lifecycle for a `follow: true` stream, not an edge case, so the `signal.aborted` early return skipping `clearTimeout` leaked one armed timer per stop. Cancelling the reader on the early loop exit closes the second half. I reformatted the changelog fragment to the `changelog.d` convention (`- **fix(scope):** …`) before merging — `check:changelog-integrity` rejects a fragment that does not start with a markdown bullet, which is the same gate your #13158 was about. Wording is yours, unchanged in substance. --- Validated in one consolidated worktree cut from `release/v3.8.51`, boarded together with the other 13 PRs of this batch — zero merge conflicts between them. - `typecheck:core` clean - complexity 2799 / baseline 3218 and cognitive-complexity 1265 / baseline 1437 — both under baseline - 71 focused assertions green across the 13 test files this batch adds or touches ⚠️ base-red inherited: #12732 — `Docs Gates (fast-path)`, `Merge integrity`, `No new ESLint warnings`, `Unit Tests fast-path` and `Fast Quality Gates` all reproduce on the pure `release/v3.8.51` tip (provider count 356 vs the 358 the modules define, SKILL.md drift, and `open-sse/utils/stream.ts` at 3115 > frozen 3098). None of them touch this diff. Thanks @anhtahaylove — the root-cause write-up, the measured before/after numbers and the red-before-green proof on every one of these made the batch reviewable as a unit.

@anhtahaylove

…outcome (#13096) `acpManager` being a module-level singleton is what turns this from a per-call leak into unbounded growth — the `MaxListenersExceededWarning` at 11 is the visible symptom. Routing every outcome through one `settle()` is the right shape, and deleting the session on the child's own exit fixes the map growth that `getActiveSessions()`'s `alive` filter was hiding. I reformatted the changelog fragment to the `changelog.d` convention (`- **fix(scope):** …`) before merging — `check:changelog-integrity` rejects a fragment that does not start with a markdown bullet, which is the same gate your #13158 was about. Wording is yours, unchanged in substance. --- Validated in one consolidated worktree cut from `release/v3.8.51`, boarded together with the other 13 PRs of this batch — zero merge conflicts between them. - `typecheck:core` clean - complexity 2799 / baseline 3218 and cognitive-complexity 1265 / baseline 1437 — both under baseline - 71 focused assertions green across the 13 test files this batch adds or touches ⚠️ base-red inherited: #12732 — `Docs Gates (fast-path)`, `Merge integrity`, `No new ESLint warnings`, `Unit Tests fast-path` and `Fast Quality Gates` all reproduce on the pure `release/v3.8.51` tip (provider count 356 vs the 358 the modules define, SKILL.md drift, and `open-sse/utils/stream.ts` at 3115 > frozen 3098). None of them touch this diff. Thanks @anhtahaylove — the root-cause write-up, the measured before/after numbers and the red-before-green proof on every one of these made the batch reviewable as a unit.

@anhtahaylove

…13100) Keeping the tail is the right direction — `sendPrompt` resolves with the stdout collected since the prompt was written and stderr is read for diagnostics after a failure, so the newest output is what callers actually use. The `[...output truncated...]` marker keeps it from being silent. Resetting `stderrBuffer` alongside `stdoutBuffer` fixes the subtler half: diagnostics for one prompt were carrying stale output from every earlier one. Verified on the tree that actually ships — your branch merged onto the current tip, which already carries #13096: `appendCapped()` and `settle()` coexist cleanly and all 7 assertions across both ACP test files pass together. I reformatted the changelog fragment to the `changelog.d` convention (`- **fix(scope):** …`) before merging — `check:changelog-integrity` rejects a fragment that does not start with a markdown bullet, which is the same gate your #13158 was about. Wording is yours, unchanged in substance. --- Validated in one consolidated worktree cut from `release/v3.8.51`, boarded together with the other 13 PRs of this batch — zero merge conflicts between them. - `typecheck:core` clean - complexity 2799 / baseline 3218 and cognitive-complexity 1265 / baseline 1437 — both under baseline - 71 focused assertions green across the 13 test files this batch adds or touches ⚠️ base-red inherited: #12732 — `Docs Gates (fast-path)`, `Merge integrity`, `No new ESLint warnings`, `Unit Tests fast-path` and `Fast Quality Gates` all reproduce on the pure `release/v3.8.51` tip (provider count 356 vs the 358 the modules define, SKILL.md drift, and `open-sse/utils/stream.ts` at 3115 > frozen 3098). None of them touch this diff. Thanks @anhtahaylove — the root-cause write-up, the measured before/after numbers and the red-before-green proof on every one of these made the batch reviewable as a unit.

@anhtahaylove

…iegosouzapw#13100) Keeping the tail is the right direction — `sendPrompt` resolves with the stdout collected since the prompt was written and stderr is read for diagnostics after a failure, so the newest output is what callers actually use. The `[...output truncated...]` marker keeps it from being silent. Resetting `stderrBuffer` alongside `stdoutBuffer` fixes the subtler half: diagnostics for one prompt were carrying stale output from every earlier one. Verified on the tree that actually ships — your branch merged onto the current tip, which already carries diegosouzapw#13096: `appendCapped()` and `settle()` coexist cleanly and all 7 assertions across both ACP test files pass together. I reformatted the changelog fragment to the `changelog.d` convention (`- **fix(scope):** …`) before merging — `check:changelog-integrity` rejects a fragment that does not start with a markdown bullet, which is the same gate your diegosouzapw#13158 was about. Wording is yours, unchanged in substance. --- Validated in one consolidated worktree cut from `release/v3.8.51`, boarded together with the other 13 PRs of this batch — zero merge conflicts between them. - `typecheck:core` clean - complexity 2799 / baseline 3218 and cognitive-complexity 1265 / baseline 1437 — both under baseline - 71 focused assertions green across the 13 test files this batch adds or touches ⚠️ base-red inherited: diegosouzapw#12732 — `Docs Gates (fast-path)`, `Merge integrity`, `No new ESLint warnings`, `Unit Tests fast-path` and `Fast Quality Gates` all reproduce on the pure `release/v3.8.51` tip (provider count 356 vs the 358 the modules define, SKILL.md drift, and `open-sse/utils/stream.ts` at 3115 > frozen 3098). None of them touch this diff. Thanks @anhtahaylove — the root-cause write-up, the measured before/after numbers and the red-before-green proof on every one of these made the batch reviewable as a unit.

@anhtahaylove

…ady aborted (diegosouzapw#13106) Correct: an `abort` listener registered on an already-aborted signal never fires, and `safeEnqueue` can't save it because enqueuing into an unread stream only buffers. The abort-later test earning its keep as a guard on the healthy path is the right instinct. I reformatted the changelog fragment to the `changelog.d` convention (`- **fix(scope):** …`) before merging — `check:changelog-integrity` rejects a fragment that does not start with a markdown bullet, which is the same gate your diegosouzapw#13158 was about. Wording is yours, unchanged in substance. --- Validated in one consolidated worktree cut from `release/v3.8.51`, boarded together with the other 13 PRs of this batch — zero merge conflicts between them. - `typecheck:core` clean - complexity 2799 / baseline 3218 and cognitive-complexity 1265 / baseline 1437 — both under baseline - 71 focused assertions green across the 13 test files this batch adds or touches ⚠️ base-red inherited: diegosouzapw#12732 — `Docs Gates (fast-path)`, `Merge integrity`, `No new ESLint warnings`, `Unit Tests fast-path` and `Fast Quality Gates` all reproduce on the pure `release/v3.8.51` tip (provider count 356 vs the 358 the modules define, SKILL.md drift, and `open-sse/utils/stream.ts` at 3115 > frozen 3098). None of them touch this diff. Thanks @anhtahaylove — the root-cause write-up, the measured before/after numbers and the red-before-green proof on every one of these made the batch reviewable as a unit.

@anhtahaylove

…iegosouzapw#13109) Right precedent — diegosouzapw#7494 fixed exactly this for the sql.js adapter and the `node:sqlite` one never got the same treatment, even though it is the default driver whenever better-sqlite3 is unavailable. `/api/db-backups/import` opening a throwaway adapter per request makes it reachable. I reformatted the changelog fragment to the `changelog.d` convention (`- **fix(scope):** …`) before merging — `check:changelog-integrity` rejects a fragment that does not start with a markdown bullet, which is the same gate your diegosouzapw#13158 was about. Wording is yours, unchanged in substance. --- Validated in one consolidated worktree cut from `release/v3.8.51`, boarded together with the other 13 PRs of this batch — zero merge conflicts between them. - `typecheck:core` clean - complexity 2799 / baseline 3218 and cognitive-complexity 1265 / baseline 1437 — both under baseline - 71 focused assertions green across the 13 test files this batch adds or touches ⚠️ base-red inherited: diegosouzapw#12732 — `Docs Gates (fast-path)`, `Merge integrity`, `No new ESLint warnings`, `Unit Tests fast-path` and `Fast Quality Gates` all reproduce on the pure `release/v3.8.51` tip (provider count 356 vs the 358 the modules define, SKILL.md drift, and `open-sse/utils/stream.ts` at 3115 > frozen 3098). None of them touch this diff. Thanks @anhtahaylove — the root-cause write-up, the measured before/after numbers and the red-before-green proof on every one of these made the batch reviewable as a unit.

@anhtahaylove

…osouzapw#13114) `stop()` is the normal lifecycle for a `follow: true` stream, not an edge case, so the `signal.aborted` early return skipping `clearTimeout` leaked one armed timer per stop. Cancelling the reader on the early loop exit closes the second half. I reformatted the changelog fragment to the `changelog.d` convention (`- **fix(scope):** …`) before merging — `check:changelog-integrity` rejects a fragment that does not start with a markdown bullet, which is the same gate your diegosouzapw#13158 was about. Wording is yours, unchanged in substance. --- Validated in one consolidated worktree cut from `release/v3.8.51`, boarded together with the other 13 PRs of this batch — zero merge conflicts between them. - `typecheck:core` clean - complexity 2799 / baseline 3218 and cognitive-complexity 1265 / baseline 1437 — both under baseline - 71 focused assertions green across the 13 test files this batch adds or touches ⚠️ base-red inherited: diegosouzapw#12732 — `Docs Gates (fast-path)`, `Merge integrity`, `No new ESLint warnings`, `Unit Tests fast-path` and `Fast Quality Gates` all reproduce on the pure `release/v3.8.51` tip (provider count 356 vs the 358 the modules define, SKILL.md drift, and `open-sse/utils/stream.ts` at 3115 > frozen 3098). None of them touch this diff. Thanks @anhtahaylove — the root-cause write-up, the measured before/after numbers and the red-before-green proof on every one of these made the batch reviewable as a unit.

@anhtahaylove

…outcome (diegosouzapw#13096) `acpManager` being a module-level singleton is what turns this from a per-call leak into unbounded growth — the `MaxListenersExceededWarning` at 11 is the visible symptom. Routing every outcome through one `settle()` is the right shape, and deleting the session on the child's own exit fixes the map growth that `getActiveSessions()`'s `alive` filter was hiding. I reformatted the changelog fragment to the `changelog.d` convention (`- **fix(scope):** …`) before merging — `check:changelog-integrity` rejects a fragment that does not start with a markdown bullet, which is the same gate your diegosouzapw#13158 was about. Wording is yours, unchanged in substance. --- Validated in one consolidated worktree cut from `release/v3.8.51`, boarded together with the other 13 PRs of this batch — zero merge conflicts between them. - `typecheck:core` clean - complexity 2799 / baseline 3218 and cognitive-complexity 1265 / baseline 1437 — both under baseline - 71 focused assertions green across the 13 test files this batch adds or touches ⚠️ base-red inherited: diegosouzapw#12732 — `Docs Gates (fast-path)`, `Merge integrity`, `No new ESLint warnings`, `Unit Tests fast-path` and `Fast Quality Gates` all reproduce on the pure `release/v3.8.51` tip (provider count 356 vs the 358 the modules define, SKILL.md drift, and `open-sse/utils/stream.ts` at 3115 > frozen 3098). None of them touch this diff. Thanks @anhtahaylove — the root-cause write-up, the measured before/after numbers and the red-before-green proof on every one of these made the batch reviewable as a unit.

@anhtahaylove

…iegosouzapw#13100) Keeping the tail is the right direction — `sendPrompt` resolves with the stdout collected since the prompt was written and stderr is read for diagnostics after a failure, so the newest output is what callers actually use. The `[...output truncated...]` marker keeps it from being silent. Resetting `stderrBuffer` alongside `stdoutBuffer` fixes the subtler half: diagnostics for one prompt were carrying stale output from every earlier one. Verified on the tree that actually ships — your branch merged onto the current tip, which already carries diegosouzapw#13096: `appendCapped()` and `settle()` coexist cleanly and all 7 assertions across both ACP test files pass together. I reformatted the changelog fragment to the `changelog.d` convention (`- **fix(scope):** …`) before merging — `check:changelog-integrity` rejects a fragment that does not start with a markdown bullet, which is the same gate your diegosouzapw#13158 was about. Wording is yours, unchanged in substance. --- Validated in one consolidated worktree cut from `release/v3.8.51`, boarded together with the other 13 PRs of this batch — zero merge conflicts between them. - `typecheck:core` clean - complexity 2799 / baseline 3218 and cognitive-complexity 1265 / baseline 1437 — both under baseline - 71 focused assertions green across the 13 test files this batch adds or touches ⚠️ base-red inherited: diegosouzapw#12732 — `Docs Gates (fast-path)`, `Merge integrity`, `No new ESLint warnings`, `Unit Tests fast-path` and `Fast Quality Gates` all reproduce on the pure `release/v3.8.51` tip (provider count 356 vs the 358 the modules define, SKILL.md drift, and `open-sse/utils/stream.ts` at 3115 > frozen 3098). None of them touch this diff. Thanks @anhtahaylove — the root-cause write-up, the measured before/after numbers and the red-before-green proof on every one of these made the batch reviewable as a unit.

docs(changelog): format the merged 13094 fragment as a markdown bullet

a1c2d97

The fragment shipped as a bare paragraph, so check-changelog-integrity fails on every branch built off release/v3.8.51, not just the PR that added it.

anhtahaylove requested a review from diegosouzapw as a code owner September 10, 2026 02:32

anhtahaylove mentioned this pull request Sep 10, 2026

docs: correct the migration count from 171 to 172 (unblocks Docs Gates) #13160

Closed

anhtahaylove mentioned this pull request Sep 10, 2026

fix(telegram): authenticate webhook deliveries with the Telegram secret token #13175

Merged

anhtahaylove closed this Sep 10, 2026

Navigation Menu

Uh oh!

docs(changelog): format the merged 13094 fragment as a markdown bullet - #13158

docs(changelog): format the merged 13094 fragment as a markdown bullet#13158
anhtahaylove wants to merge 1 commit into
diegosouzapw:release/v3.8.51from
huuhungn:fix/changelog-13094-fragment-format

anhtahaylove commented Sep 10, 2026

Uh oh!

anhtahaylove commented Sep 10, 2026

Uh oh!

anhtahaylove commented Sep 10, 2026

Uh oh!

anhtahaylove commented Sep 10, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

anhtahaylove commented Sep 10, 2026

Uh oh!

anhtahaylove commented Sep 10, 2026

Uh oh!

anhtahaylove commented Sep 10, 2026

Review queue: 14 open PRs from a resource-leak audit

Please start here — these two are one-line docs fixes that unblock everything else

The twelve leak fixes

On the red CI you'll see

Uh oh!

anhtahaylove commented Sep 10, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant