Skip to content

fix: 0.12.2 batch 5 (7 fixes from main) - #34923

Merged
bolichen97 merged 7 commits into
release/0.12.0from
hotfix/0.12.2-batch5
Nov 11, 2026
Merged

bolichen97 merged 7 commits into
release/0.12.0from
hotfix/0.12.2-batch5

Conversation

@bolichen97

@bolichen97 bolichen97 commented Nov 11, 2026 •

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: Ship seven more of our own fixes from main to 0.12 users in 0.12.2, on release/0.12.0.

These fix bugs that are in release code. Each is adapted to release's code and leaves out what needs code only on main; every adaptation is a Backport note: line in its commit.

Why it matters

0.12 users hit each of these today: a cron run cut by a restart that is reported as nothing, a KAS MCP sign-in saved where KAS chats do not read it, activity-index rows stamped instead of listed, a kiro-cli engine refusal that is not treated as a host verdict, a worktree create refusal with no reason, KAS sessions that start in the wrong mode or aim sub-agents at the wrong target, and a dead pooled MCP backend that is not named.

Not a goal

  • No new behaviour on release: each commit is a backport of the main fix; a hand-port keeps release's own code and says what it left out in a Backport note: line.
  • The core update-restart fix, and any other open PR, is not in this batch.

What changed (motivation → approach → change)

The 7 picks, in main merge order, on release/0.12.0 at a7a6dd4397:

# PR Main merge commit Fix Restaged
1 #33905 ffc89b7979 report a restart-cut cron run and send call_tool args as written no
2 #33737 c84e4d728e save a KAS MCP sign-in where KAS chats read it no
3 #33982 f4b45992f0 list history entries in the activity index, not their stamps no
4 #34004 4febed4cfc treat kiro-cli's v3 engine refusal as a host verdict no
5 #34057 b1be19dddf name git's reason when worktree create refuses a project no
6 #33215 f6bc8875f9 four KAS session start, mode and sub-agent target fixes yes, the PR head moved after staging
7 #33595 35dc109f0d name a dead pooled backend and an unresolved core ref yes, the PR head moved after staging

Hand-ports with a Backport note: carry what each commit body lists as adapted or left out.

Backwards compatibility

No behaviour change beyond each listed fix. No config key, CLI flag or API is removed or renamed.

Tests

On a DevSpace (Python 3.12 venv, node 22.22.1), nothing run on a laptop. release/0.12.0 has no backend CI on pull requests (ci.yml targets main), so these runs are the gate:

  • Union of the seven picks' related test files (touched tests, test_<module>* siblings, and tests that name a touched module): 1066 files in 18 batches with pytest -n auto --dist loadgroup --max-worker-restart=2. 39677 passed, 103 skipped, 1 xfailed, 12 failed. All 12 failures also fail on plain release/0.12.0 (a7a6dd4) in the same host: 10 in test_cli_stt_memory_doctor.py (stale timezone data on the host makes doctor exit 1), test_gateway_lock_diagnosis.py::test_flock_is_held_by_a_fork_orphan, and test_kas_cli_floor.py::...kills_the_whole_probe_tree (os.pidfd_open absent). 20 helper, e2e and integration files in the related set were not run; the gateway smoke below covers boot.
  • Import check of all 75 touched modules: 75 imported, 0 failed.
  • Website: tsc -p tsconfig.app.json clean, npm run build ok, 18 vitest files (374 tests) for the touched website code passed, after npm ci.
  • Smoke: wheel built with the staged dashboard, installed in a fresh venv: kirocrew --version prints 0.12.1; kirocrew doctor runs (same host-state findings as on plain release); a gateway on a temp KIROCREW_HOME and a free port answered 200 on / and /api/health, then stopped by PID.

Rebased once onto the new release/0.12.0 tip a157d53d25 (three KAS PRs landed during the run; the rebase was clean, 8 files overlap). Re-run on the rebased tree: the related tests of both sets plus the sibling tests of the 8 overlapping files, 183 files: 8538 passed, 62 skipped, 1 failed (the same os.pidfd_open host failure that fails on plain release); import check of 92 modules, 0 failed; wheel install and gateway smoke: / and /api/health 200, stopped by PID.

Proof: not provable -- other: a backport batch of already-merged main fixes, each proven in its own main PR

Manual verification

N/A: each fix was verified in its own main PR; the batch gets a wheel install and gateway smoke on a DevSpace (see Tests).

Screenshots / video

N/A: no visual change.

Related Issues

0.12.2 hotfix, batch 5: #33905, #33737, #33982, #34004, #34057, #33215, #33595.

Sibling sweep

Siblings: none -- a backport batch; each pick's own PR carried its sweep

Pattern harvest

Not generalizable: a release backport batch, not a new defect fix.

Checklist

  • Each commit is a backport of a main commit with a Conventional Commits title
  • Existing tests pass and new tests added for new functionality
  • Self-review completed; code follows project style guidelines
  • Documentation updated (if applicable)
  • No secrets, credentials, or internal references in the diff

Contribution License Agreement

Contributions are accepted under the project's Apache-2.0 license; see the Licensing section of CONTRIBUTING.md.

@bolichen97
bolichen97 requested review from a team and KiroCrewCaptain as code owners November 11, 2026 00:26
@bolichen97
bolichen97 requested review from dodgeblaster and removed request for a team November 11, 2026 00:26
@github-actions

github-actions Bot commented Nov 11, 2026 •

Copy link
Copy Markdown
Contributor

Sensitive Change Review

Sensitive files changed:

  • src/kiro_crew/security_posture_sinks_gateway.py

  • @buluoray: missing the ## Sensitive change review heading

How to clear this check:

  1. The PR author posts the template below as a PR comment, with
    every field filled (N/A needs a reason). Keep the
    Head commit line: it must name the current head SHA.
  2. The check turns green within a minute, with no push.
  3. A new push resets this check: post it again for the new head.

Only the author's comment counts this way. A reviewer can instead
paste the filled template into an "Approve" review on this head.
Merging still needs a reviewer's approval.

## Sensitive change review
- Head commit: cd1a83cdc338d2f5bdcfcd985711feea6fd809bd
- What rule changed:
- Before -> after (what was allowed/blocked, now):
- Worst case if wrong (breaks commands / leaks secret / locks user out):
- Evidence checked (which sandbox result proves it):
- No regression:
  - Tools still work (which tools/commands were run, result):
  - Backward compatible (old config / old data / old callers still work, how checked):
  - Existing tests (which suites passed on this head):
- How to undo:

@github-actions github-actions Bot added the readiness: checking Automated validation is still running label Nov 11, 2026
@github-actions

Copy link
Copy Markdown
Contributor

First Principles Review (Opus 5.5) — ✅ PASS

Premise-level review of 6b3ea1570abf90b45f6b59e72ce612a3fbccc74c — why this exists and whether the shipped surface is the smallest honest version. Updated in place on each push. A BLOCK verdict blocks PR readiness; PASS does not.

First-Principles-Verdict: PASS

This PR backports seven fixes that are already merged on main. Each one comes from a linked main PR that has its own provenance and sibling sweep. No item changes a default or adds new surface beyond its fix.

What this change ships

Inventory (8 items) — 8 justified

Intent: bring seven already-merged main fixes to 0.12 users in 0.12.2. This is a FIX.

  1. kirocrew restart and an update restart now name the cron runs they cut short, record them as cancelled, and say whether each job runs again — justified
  2. A follow-up suggestion's call_tool args are now sent exactly as written — justified
  3. A KAS MCP sign-in made from the MCP or Connections row is now saved where KAS chats read it, and the KAS chat note is now set per row — justified
  4. Activity-index rows now list history entries instead of their stamps — justified
  5. A kiro-cli refusal of its v3 engine is now recorded as a host verdict and read before a KAS config load — justified
  6. A refused worktree create now gives git's reason as a code and a redacted detail — justified
  7. KAS sessions now start in the correct mode, and their sub-agents point at the correct targets — justified
  8. A dead pooled MCP backend, and an unresolved core ref, are now named — justified

[FIRST-PRINCIPLES-REVIEWED] 6b3ea15

Description read

[DESCRIPTION-READ] 63987ffa1b62dc1977745e1bfc24bf0e4e7a3280d824f224c5fdf4de279d1b43

That sha256 names the PR description this verdict read. Recompute it with .github/scripts/pr-description-capture.sh; if it differs, the description changed after this verdict was formed and any finding drawn from the description is unproven.

@github-actions

Copy link
Copy Markdown
Contributor

GPT 6.1 Review — 🔴 changes requested (blocking)

GPT 6.1 found at least one blocking issue that must be resolved before merging 6b3ea1570abf90b45f6b59e72ce612a3fbccc74c. Adjudication upheld 1 of 1 blocking finding(s); those must be resolved before merging.

This comment is updated in place on each push.

BLOCKING -- src/kiro_crew/slack/gateway_runtime/cron_failure.py:394 -- Paused jobs incorrectly promise another scheduled run

if job.schedule.kind != "at":
text += " The job runs again at its next scheduled time"
Trigger: Run a paused recurring job with kirocrew cron trigger <id>, then restart during execution. The CLI calls the run API, which accepts disabled jobs; stop() cancels the run and invokes _alert_cron_interrupted().
Outcome: The notification promises an automatic rerun, but the result merge preserves the pause and the scheduler excludes disabled jobs. The promised run never occurs unless the user resumes or manually triggers it.
Anchor: defect
Fix: Append the next-run promise only when the post-merge stored job remains enabled; otherwise report just the interruption.

[BLOCK-MERGE] 6b3ea15
[GPT-REVIEWED] 6b3ea15

Adjudication (Opus 5.5) — is blocking on each finding proportionate?

F1: LOW harm. The message wrongly promises an automatic rerun. It only appears when someone manually triggers a paused recurring job and the gateway restarts while it runs; the user still sees the job as paused. The message comes from src/kiro_crew/slack/gateway_runtime/cron_failure.py:394. It is sent through _alert_interrupt (src/kiro_crew/cron_service/service/run_execution.py:341-349), and the scheduler skips disabled jobs (src/kiro_crew/cron_service/service/wake_timer.py:178), so the promised run never happens. Nothing corrects the message afterwards.
The real fix is cheap: one more condition, only adding the rerun sentence when the job is still enabled. There is no new state and no coordination between modules, so the fix's cost doesn't outweigh even this low harm.

[ADJUDICATION] 6b3ea1570abf90b45f6b59e72ce612a3fbccc74c total=1 uphold=1 downgrade=0
UPHOLD F1 src/kiro_crew/slack/gateway_runtime/cron_failure.py:394 reason=harm-warrants-remedy
[GPT-ADJUDICATED] 6b3ea1570abf90b45f6b59e72ce612a3fbccc74c

False positive or not applicable? A repository writer can comment:
/ai-review override gpt 6b3ea1570abf90b45f6b59e72ce612a3fbccc74c false-positive: <reason, 20 to 500 characters>

@github-actions github-actions Bot added readiness: action required A blocking check or review needs attention and removed readiness: checking Automated validation is still running labels Nov 11, 2026

@buluoray buluoray left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Approve on 6b3ea15. Seven backports of KAS/cron/connections/dashboard/MCP fixes. Every main twin is merged and on origin/main. Five picks are hand-resolved and each keeps its main fix's intent. No duplicates. No test failures caused by the batch. The PR is based on a7a6dd4, and release has since moved to a157d53 (#31892, #33045, #33478). merge-tree with live release exits 0, and the KAS suites pass on the merged tree. Decisive lines: GPT 6.1's upheld blocker at src/kiro_crew/slack/gateway_runtime/cron_failure.py:394 is real but inherited byte-for-byte from main ffc89b7 (#33905). The pick's src patch is identical to main, and origin/main has the same line at :394-395. No main fix exists yet, so under main-first it is a main follow-up, not a release blocker. The GPT lane needs a writer override.

Backport of #33905 (main ffc89b7). Conflicts resolved to the release side: dropped main-only trigger_run, expected_owner and overlay_owners imports and the quiet_run context lines; added set_interrupt_alert only. security.md: the PR sentence was applied into the release JWT bullet. test_cron_stop_interrupt_notice.py got a local _trigger_run helper in place of CronService.trigger_run.
Backport of #33737 (head 7d13e5f).
Conflicts resolved on the release side with the crew_oauth parts dropped (absent on release); crewHeld is carried for the KAS sign-in only; the pid-reader owner-marked baseline is 12; the McpTab crew-held tests are dropped.
Backport of #33982 (head 153d189). Applied clean; no conflicts, no dropped parts, no carried helpers.
Backport of #34004 (head b9596d4).
Release names the LTS gate enterprise_boot (renamed to lts_boot on main), so the lts_boot.py and test_lts_harness_floor.py hunks were applied to enterprise_boot.py and test_enterprise_harness_floor.py with LTS wording changed to enterprise. Conflicts in platform/bootstrap.py and providers/acp_runtime_start.py were import-line context only (main-only bots_may_run_on_kas not carried). harness-parity.md: H1 row edits re-applied by hand, one Pinned-by test reference whose neighbour is absent on release was dropped.
Backport of #34057 (head d26753a).
Docs: kept release text and applied only this PR's wording (the separate-copy table row is absent on release, so dropped); error-code-baseline.json regenerated on release (missing_code 968 to 967, worktree.py 16 to 15).
Backport of #33215 (head 006da76 (merged as main f6bc887; restaged onto the final PR head)).

Carried the session_request seam (base default plus the create_session
call) the mode fix needs; dropped the Kiro Cloud doc hunk, the cloud
test, governance_app and the cloud-only reader seams.
Backport of #33595 (head a73e4da (merged as main 35dc109; restaged onto the final PR head)).

Conflict resolution: acp-client.md module table, kept release rows and
added only the PR text on the preparation.py row.
@bolichen97
bolichen97 merged commit 68403ee into release/0.12.0 Nov 11, 2026
17 of 34 checks passed
@bolichen97
bolichen97 deleted the hotfix/0.12.2-batch5 branch November 11, 2026 01:58
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

readiness: action required A blocking check or review needs attention

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants