Repository navigation
fix: backport 0.12.1 fast batch v2b-2 (16 user-facing fixes) - #32977
Conversation
…31046) _set_default_agent and the Slack users and channels select handlers set one key inside the agent or slack config section. When that entry held a list, string, number or null, setdefault returned it and the item assignment raised TypeError, an unhandled crash out of the locked write. Each mutate now returns None for a non-object section, so nothing is written and runtime state is unchanged. The select handlers log and return as on an unreadable config, and _set_default_agent raises ValueError. The /kirocrew agent slash handler now replies with that error, as the !agent command does. (cherry picked from commit 6792d17)
Co-authored-by: w <w@k.local> (cherry picked from commit 4a35d06)
…tarting (#22017) asyncio.create_subprocess_exec forks the child before it finishes connecting the pipes, so a cancel inside that await went through asyncio's own transport cleanup. That cleanup sends SIGKILL to the leader pid alone. CommandProvider.apply never held the process, and the installer's TERM trap never ran, leaving the venv moved aside and the rest of its group alive. The child now starts in a task the cancel does not interrupt. A cancel that arrives during the start waits for it inside the TERM grace and then stops the child through _stop_installer with the grace that is left, so the whole stop still fits UPDATE_INSTALLER_STOP_SECS. A start that ends after the grace has run out is logged, since that child gets SIGKILL with no graceful stop. A start still pending when the grace runs out is cancelled, which falls back to asyncio's own kill. A regression test holds the pipe connection open to force the window, and a second test holds it past the grace. Closes #21882 (cherry picked from commit 89a6107)
(cherry picked from commit 4c55f55)
Co-authored-by: w <w@k.local> (cherry picked from commit 0140686)
Every streamed chunk made a new transcript, and the whole transcript went through the slot-scoped deferred value, so each chunk scheduled a background render of the chat page. On a slow phone the next chunk interrupted that render every time, so the reply's text almost never updated; on desktop it cost a second page render per chunk. A chunk that only grows the last row renders at urgent priority: useSlotDeferredValue takes an optional sameShape check, and useTranscriptRows passes grewOnlyAtTail. A landed page, a new row or a regroup is still deferred, and while one is pending the hook shows the last value it rendered live, never an older frame. The window while a change of shape is pending is tracked in #31704. Builds on #23487. Closes #23486 Co-authored-by: Pearcekieser <5055971+Pearcekieser@users.noreply.github.com> (cherry picked from commit 9d37818)
(cherry picked from commit d939b8c)
…31445) The artifact-deploy skill's pre-upload scans (deploy.sh, _common.sh) and the ops ledger's pre-push scan ran get_credential_patterns()'s raw search over whole files or lines. That search re-walks a long eyJ or digit run from every start, so one such file or line stalls the scan for seconds to minutes. Export the shared linear scanner as kiro_crew.security's public iter_credential_matches and move every caller outside the package to it, including the two that imported the private _credential_matches. Drop the unused get_credential_patterns re-export from the design tweak server. Findings stay those redact_credentials sees, and a scanner error still aborts the deploy. (cherry picked from commit 7aaf1b1)
A guide replay ("Show again") takes a completed guide back to offered,
patching the same transcript row in place. The slot-detail merge could
not tell a stale snapshot of that row from a newer one, so a snapshot
taken before a patch redrew the card's earlier status in either
direction.
Each in-place patch of a card row now bumps meta.card.rev, and the
dashboard merge keeps the copy with the higher revision whichever way
the status moved. Copies with equal revisions keep the existing
status-anchor rule.
Closes #22795
(cherry picked from commit e702910)
…before a response (#27055) An SSM port-forward can accept the connection and drop the stream before the peer answers; aiohttp raises ServerDisconnectedError and the dashboard shows a one-off 502 proxy_peer_unreachable while the next dial succeeds. GET/HEAD/OPTIONS now get one fresh dial for a drop or reset; refused dials, timeouts and state-changing methods are not redialled. (cherry picked from commit f09928e)
…32370) POST /api/update-message rewrites one of the bot's own Slack posts with new agent-authored text, but it never asked the calling app's own capabilities.messaging grant. Every send_message egress leg asks it at the gateway, because the MCP tool's vet runs where no app is known. An app whose messaging is revoked could still reach Slack by editing a message it posted earlier. Ask _vet_app_messaging before the target is resolved, audited under update_message, and answer 403 channel_not_permitted like a refused send. Co-authored-by: Leon <leon.lai.yc@gmail.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> (cherry picked from commit 3d72fae)
AppManifest.from_dict read mcpServers as dict(data.get("mcpServers",
{})). A genuine non-object made that dict() raise -- TypeError on a
scalar or bool, ValueError on a non-empty list or string -- and the
exception escaped install, discovery and lookup as an uncaught 500
(install catches ValueError but not TypeError). Wrap it so a non-object
raises one clear ValueError instead.
The empty forms dict() already accepted ([] and "", which yield {}) keep
parsing to an empty map, so no manifest that parses today is newly
rejected. The raise keeps every reader that skips validate() fail-closed
exactly as before: get_app_manifest returns None on a parse error, so
the dashboard MCP-ownership PUT and the boot reconcile prune treat the
manifest as unreadable and preserve the app's live bridges rather than
reading an empty map as "declares none". The install path surfaces the
ValueError as a clean refusal.
Fixes #31178
(cherry picked from commit 359db48)
Sensitive Change ReviewSensitive files changed:
How to clear this check:
Only the author's comment counts this way. A reviewer can instead |
UX Review (Opus 5.5) — ✅ PASS (a BLOCK with no Trigger and no Outcome does not block)UX-level review of UX-Verdict: PASS cannot evaluate: missing screenshots of the new theme-delete dialog and the muted crew stall states. The PR changes visible UI but attaches no images, so the blind read was not performed. Evidence gaps
Notes
[UX-REVIEWED] 0bcd8fa Description read[DESCRIPTION-READ] f062a0d508ce2c8dd75a91c8c13ef761923701ebc08532970d17e4ba0d860712 That sha256 names everything this verdict read: the PR description, plus the 3 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
Design Review (Opus 5.5) — ✅ PASS (a BLOCK with no Trigger and no Outcome does not block)Design-level review of Design-Verdict: PASS cannot evaluate: missing a screenshot of the in-app theme-delete dialog and the crew sidebar's new red/muted states, rendered from this revision. The evidence list names nothing, and the description points only to the main PRs' screenshots. BlockersThe new or changed UI surfaces have no rendered evidence for this revision Notes
[DESIGN-REVIEWED] 0bcd8fa Description read[DESCRIPTION-READ] e152f43465b0cb99fdc7e765cfc7fdf49591c500cfe4c1573019893bf20f5ec3 That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
First Principles Review (Opus 5.5) — ✅ PASSPremise-level review of First-Principles-Verdict: PASS Nothing blocks: every item is a clean cherry-pick of a fix already merged on What this change shipsInventory (10 items) — 10 justifiedIntent: bring sixteen fixes already merged on
[FIRST-PRINCIPLES-REVIEWED] 0bcd8fa Description read[DESCRIPTION-READ] 53d45c0afaff00709806f05040d87a66b0e274d1049a5668fb142e7f6fc60631 That sha256 names the PR description this verdict read. Recompute it with |
Security Scope Review -- ✅ PASSWhich legitimate operations does Verdict basis: adjudicated, zero confirmed regressions. Security scope review -- adjudicated candidates
No candidate flipped from allowed to refused. Scope-Verdict: PASS Reason: The patch does not touch the shell-command deny checks ( TIGHTENING: The main one is the live removal of ADJUDICATED: 15 candidates, 0 confirmed newly-refused PLATFORM COVERAGE: posix (11 rows) covers the operations maintainers run in the touched areas: the new and neighbouring tests, UNADJUDICATED:
[SCOPE-REVIEWED] 0bcd8fa |
GPT 6.1 Review — ✅ no blocking findingsGPT 6.1 completed its review of This comment is updated in place on each push. Review detailsNo findings. False positive or not applicable? A repository writer can comment: |
Opus 5.5 Review — ✅ no blocking findingsReviewed Verdict parsed from the review's SHA-scoped output markers for commit False positive or not applicable? A repository writer can comment: |
Problem / Motivation
Goal: Bring sixteen fixes for problems a 0.12.0 user can hit, merged on
mainafterv0.12.0, ontorelease/0.12.0, one commit each, in a PR of its own.Why it matters
Each fix changes behaviour a 0.12.0 user can run into today. Every pick applied to
release/0.12.0cleanly, so the code each one replaces is on release.Not a goal
main; nothing is changed only here.CHANGELOG.mdchange.What changed (motivation → approach → change)
Each main squash commit was cherry-picked with
-xontorelease/0.12.0@b447cd4fc6, in merge order onmain, one commit each. Every pick applied cleanly, and every commit keeps its original author.6792d171e4b0268785f37e339ced37ac1e43b5ca4a35d06f114514805c8b89a61071c536a7cc186b1ed5cf491dc17249aec84c55f55efbecc9ac94b90140686ccd171ecedd712b0c1350592df3292c8e9d37818b2a0d00995fc9d939b8c16832888a68fc7aaf1b10d7d9b91ebde6e7029108068baa1506d2f09928e823f4bc5696c13d72faec34c2f37eedcd9aeb2ebb92160587a550359db48c030bcd8fa9d9Picks whose co-author trailers carried internal addresses have each trailer rewritten as a plain "Co-author on main: " line, because the commit guard refuses the address.
#30296, #31018, #30516 and #31949 now conflict, in a test, translator notes, a test and a spec doc only, so they wait for a docs/tests batch.
Backwards compatibility
Compatible: each change is the one already merged on
main, and its compatibility note applies unchanged.Removes nothing: no setting, key or caller is removed.
Tests
Proof: not provable -- other: a backport of main commits whose proofs ran on their own PRs
release/*, so the direct test modules were run here: the 22 Python test files these picks touch gave 1,646 passed and 3 skipped.tsc -p tsconfig.app.json --noEmitpasses; the 10 vitest files these picks touch gave 285 passed;lint:i18npasses at the same count as clean release.mypy src/kiro_crewpass.i18n:checkwithout a base reports the same findings class as a cleanrelease/0.12.0on this host, and fewer of them.Manual verification
N/A - each fix was verified on its own main PR. The 0.12.1 RC soak is the end-to-end check.
Screenshots / video
Why no screenshot: the frontend picks carry the same rendered change as their main PRs, which show it.
Related Issues
Backport to the 0.12 line for 0.12.1 of #31046, #31092, #30840, #22017, #30991, #31777, #31983, #31810, #31719, #31703, #31445, #32325, #27055, #32370, #32502, #31194.
Sibling sweep
Siblings: none -- each source PR carried its own sibling sweep; this PR only carries those fixes.
Pattern harvest
Not generalizable: this PR only carries fixes already on
main. The source PRs have their own harvest.Checklist
feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)Contribution License Agreement
Contributions are accepted under the project's Apache-2.0 license; see the Licensing section of CONTRIBUTING.md.