Skip to content

fix(release): backport ten user-reported fixes to 0.12.0 line - #31968

Merged
buluoray merged 10 commits into
release/0.12.0from
crew/slack/0.12.1-batch-20
Nov 7, 2026
Merged

buluoray merged 10 commits into
release/0.12.0from
crew/slack/0.12.1-batch-20

Conversation

@buluoray

@buluoray buluoray commented Nov 7, 2026

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: Bring ten small fixes for user-reported issues that merged on main after v0.12.0 onto release/0.12.0, one commit per fix, so the 0.12.1 patch ships them and each one can be reverted alone.

Why it matters

PR Reported in What a 0.12.0 user sees today
#30258 #30191 (cschnidr) An update whose pre-apply fetch takes over 30 s fails with a 500, though the pull itself allows 60 s.
#30500 #15220 (jongaffen) A partial subagent digest gives a still-running count with no time on it.
#30498 #14762 (tmack8001) Back or forward onto a closed session lands silently, with no word that it is closed.
#30551 #16629 (leonlaiyc) The gateway's own Python environment leaks into the AWS (Polly) child processes it starts.
#30566 #16890 (huangzhongzhong) A path written with CJK punctuation around it does not become a file chip.
#30611 #7822 (timwukp) The EC2 cloud bootstrap dies minutes in when the VPC has a private zone that shadows the kiro-cli download host.
#30460 #10583 (martinsjavacode) An app with a Node backend whose package.json is not at the app root installs no npm dependencies.
#30458 #11078 (davihara) A failed app install script leaves nothing of its output in the gateway log.
#30948 #25813 (mbriones98) A side question's text vanishes while its cancel is still in flight.
#31151 #27882 (jongaffen) The deployment list on a host that cannot run it shows no reason.

Not a goal

  • No fix that is not already merged on main; nothing is changed only here.
  • No version bump, no tag, no CHANGELOG.md change. Cutting v0.12.1-rc.N and the bare tag follow docs/build/release.md and stay with the maintainers.
  • No pick that touches auth, the sandbox, the hook or permission gate, credential handling, or shared boot or session paths. fix: name the missing credential helper in cloud doctor on windows #31205 (name the missing Windows credential helper in cloud doctor) was left out for that reason: it reads the AWS credential_process setting.

What changed (motivation → approach → change)

Each main squash commit was cherry-picked with -x onto release/0.12.0 @ c8dedc263a, in merge order, one commit each, keeping its original author.

# PR main commit release commit Pick
1 #30258 match the pre-apply fetch deadline to the pull, 409 on timeout 8b96d2ad93 ad8e4b5d55 clean
2 #30500 date the still-running count in partial subagent digests a68f38ec15 4847d2f1d5 clean
3 #30498 tell the user when back or forward lands on a closed session 91c35d326e 864a5e179d clean
4 #30551 keep the gateway's Python env out of Polly's AWS children b1f4c11d33 536be9d1bc clean
5 #30566 let paths with CJK punctuation become file chips a0a29308f2 4ef4f3e79c clean
6 #30611 download kiro-cli from a host no VPC endpoint can shadow 3ae48d99c8 51fbbe8de2 clean
7 #30460 install npm deps where a Node backend's package.json lives 087043e1f8 58f4ee9408 clean
8 #30458 keep a failed install script's output in the gateway log 1e51712ee8 3e580a3721 clean
9 #30948 keep a cancelled side question's text while its cancel waits d583482fe7 b20f91075e clean
10 #31151 show why the deployment list cannot run on this host d565591d02 76d5fbf314 clean

No hunk was resolved by hand. Every pick's added and removed lines are identical to its main commit. #30551 uses sandbox_spawn.without_python_env, which already exists on this branch; it does not change the sandbox.

builder-security checked: no pick changes a bind, a listener, an auth check, an ingress rule or a storage grant. #30258's handler change raises the pre-apply fetch timeout from 30 s to 60 s and answers a timeout with 409 instead of 500. #30611 is a supply-chain change in the tightening direction: the EC2 bootstrap now fetches the pinned kiro-cli 2.27.1 from prod.download.cli.kiro.dev instead of latest from desktop-release.q.us-east-1.amazonaws.com, and verifies it with sha256sum -c against two checksums it adds to packaging/kiro-cli-sha256. 2.27.1 is the version this branch already pins in packaging/kiro-cli-version.

Backwards compatibility

Compatible: each change is the one already merged on main, and each PR's own compatibility note applies unchanged. A client that read #30258's timeout as a 500 now gets a 409 with a retry hint.

Removes nothing: no pick hides, deletes or renames anything 0.12.0 accepts.

Tests

Proof: not provable -- other: a backport of main commits whose proofs ran on their own PRs

  • CI does not run the backend suite on PRs into release/*, so the related set was run here: every touched test file plus every test file that imports a touched module, 182 files, one pytest process per batch of 20, -n 0. 11622 passed, none failed.
  • Frontend: every touched test file plus every src/i18n/*.test.ts* catalog test, 62 files, 1421 passed.
  • flake8 passes on the changed Python files.

Manual verification

N/A - every pick was verified on its own main PR. The 0.12.1 RC soak is the end-to-end check.

Screenshots / video

Why no screenshot: each frontend change here is applied from main unchanged; its look is reviewed on its own PR, not here.

Related Issues

Backport to the 0.12 line for 0.12.1 of #30258, #30500, #30498, #30551, #30566, #30611, #30460, #30458, #30948 and #31151. Source issues: #30191, #15220, #14762, #16629, #16890, #7822, #10583, #11078, #25813, #27882.

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. Each source PR has its own harvest.

Checklist

  • At most two commits (one is the norm), with a Conventional Commits title (feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)
  • 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.

chenmingwei23 and others added 10 commits November 7, 2026 01:17
…e a slow fetch as 409 (#30258)

The dashboard Update handler fetched the tracked remote under a 30s
deadline, half the 60s the fast-forward that follows it gets, and
reported a timeout as HTTP 500. A reachable-but-slow fetch (a large
pack over a thin link) was killed and surfaced as a server error.

Raise the fetch deadline to 60s to match the fast-forward's own bound,
and answer a genuine overrun with 409 (retryable) like the adjacent
non-zero-rc refusal, with a clear reason. The fetch still runs and the
failure stays surfaced; code git_fetch_failed is unchanged so the UI
keeps showing the reason.

Closes #30191

Co-authored-by: kirocrew-agent <agent@kirocrew.local>
(cherry picked from commit 8b96d2a)
A mid-wave digest is composed when its chunk fires, with the number of
members running at that moment, and can then wait in a busy parent's
queue while those members finish. The parent read "1 still running"
for a member that had already ended. Each mid-wave digest now dates its
counts in a sentence right after the header; the header line, which a
channel parent's meta-less rows are parsed by, the spawn guidance and
the card meta are unchanged.

Closes #15220

(cherry picked from commit a68f38e)
…0498)

Back or Forward onto the history entry of a session this tab saw
closed rewrote the entry to the session on screen and said nothing,
so the enabled arrow looked dead. The chat page's history reader now
raises the pane notice with a new `closed` kind ("That session was
closed. Find its transcript in Older Sessions.") on both of those
repairs, and clears it when the next step lands on a live session.

Builds on #14765 by @tmack8001.

Closes #14762

(cherry picked from commit 91c35d3)
aws-cli v1 is a Python program. The Polly consent probe (through
cloud.aws.run_aws) and Polly synthesis handed it the gateway's
PYTHONPATH/PYTHONHOME, so on such a host the voice list worked while
consent and synthesis failed. Both children now get their environment
through sandbox.without_python_env, as the voice catalogue already does;
run_aws keeps aws_spawn_env's widened PATH.

Builds on #14207 by @leonlaiyc.

Co-authored-by: Leon <leon.lai.yc@gmail.com>
(cherry picked from commit b1f4c11)
An existing path whose name carries non-ASCII punctuation or a symbol
(the middle dot in a transliterated name, an ideographic comma, a
fullwidth plus or bracket) failed the path-shape pre-filter, so the
stat probe never ran and the chip stayed plain inline code.

The shape now admits every non-ASCII punctuation mark and symbol inside
a segment, on both the POSIX and the drive-rooted shape, and lets a
closing bracket of any script end a path. The ASCII exclusions that keep
a command or URL from classifying are unchanged, and a trailing
sentence mark such as an ideographic full stop still cannot end one.

Closes #16890

(cherry picked from commit a0a2930)
…30611)

The EC2 bootstrap fetched kiro-cli from
desktop-release.q.us-east-1.amazonaws.com, a name inside the private DNS
zone an Amazon Q interface VPC endpoint creates. In a VPC carrying that
endpoint the lookup returned NXDOMAIN and every launch failed.

The bootstrap now fetches the musl zip of the packaging/kiro-cli-version
release from prod.download.cli.kiro.dev, the host the official installer
uses, and checks it against its packaging/kiro-cli-sha256 line before
unzipping. The DNS preflight checks the new host, and its error no
longer blames an AWS endpoint.

(cherry picked from commit 3ae48d9)
…30458)

When an app's setup.onInstall script exits non-zero during a registry
install or update, the output that says why lived only in the install
response. Once that response was gone, all that remained was a generic
"install script failed (exit N)" event.

The install now also writes that output tail to the gateway log as one
warning line: redacted off the event loop through the context-aware log
redactor, then cut to a fixed size, and quoted so a newline in the
output cannot start a new log line.

Closes #11078

(cherry picked from commit 1e51712)
…0948)

A Side Chat cancel hands the question back to the composer when its
answer lands. If a socket re-read dropped the card first, the text
waits in a record that is capped at 50 cards and replaced on every
close, so more than 50 dropped cards, or a second conversation opened
and closed before the answer, threw the question away.

The tab now marks each cancel it sends as waiting until the answer
settles, and the record keeps a waiting card's text through the cap
and through later closes.

(cherry picked from commit d583482)
…1151)

A gateway that cannot list deployments at all (native Windows) answers
GET /api/deploy/list with HTTP 200 and its reason in `error`. The Deploy
page never read that field, so it showed "No deployments yet" and a hint
to wait a couple of minutes, hiding the reason the gateway sent.

The page now shows that reason through the Deployments section's
existing error notice and skips the empty state and the tag-delay hint
while it is set, as the Publish panel already does.

Closes #27882

(cherry picked from commit d565591)
@buluoray
buluoray requested review from a team and KiroCrewCaptain as code owners November 7, 2026 01:32
@buluoray
buluoray requested review from tim-finnigan and removed request for a team November 7, 2026 01:32
@github-actions github-actions Bot added the readiness: checking Automated validation is still running label Nov 7, 2026
@github-actions

github-actions Bot commented Nov 7, 2026

Copy link
Copy Markdown
Contributor

UX Review (Opus 5.5) — ✅ PASS (a BLOCK with no Trigger and no Outcome does not block)

UX-level review of 76d5fbf314f2c0c9f5c7995b31a0a071debe7cc8 — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS does not.

UX-Verdict: PASS

cannot evaluate: missing screenshots of the closed-session notice and the deployment-list reason. The PR attaches no image, so the blind read did not run.

Evidence gaps

  • Closed-session pane notice (setSwitchSlotGone({ name: '', kind: 'closed' }), new string "That session was closed. Find its transcript in Older Sessions."): no screenshot shows the chat pane after Back or Forward lands on a closed session. A screenshot of that pane would close the gap.
  • Deployment-list reason: the ErrorNotice now shows listError, and the empty state is hidden when listError is set. No screenshot shows the Deploy page on a host that cannot list deployments, such as native Windows. A screenshot of that page would close the gap.
  • Cancelled side question: noteSideQueueCancelWaiting keeps the question's text on screen while its cancel is still waiting. No screenshot shows this. A screenshot of a side chat during a pending cancel would close the gap.

Notes

[UX-REVIEWED] 76d5fbf

Description read

[DESCRIPTION-READ] dcbf9deb3c5f7da224a0ac79006ee4103ca179d5c798bc8ea655c8a0eb72ddf9

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 .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions

github-actions Bot commented Nov 7, 2026

Copy link
Copy Markdown
Contributor

Design Review (Opus 5.5) — ✅ PASS

Design-level review of 76d5fbf314f2c0c9f5c7995b31a0a071debe7cc8 — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS does not.

Design-Verdict: PASS

No design-level concerns: these are unchanged cherry-picks of fixes already merged on main. The new UI text reuses the existing notice components and adds no new page, panel or control set.

Notes

[DESIGN-REVIEWED] 76d5fbf

Description read

[DESCRIPTION-READ] cfc39afc2acac162d3dd4c31ff8728f87fb9763a75b0220cc1f9beeb3387f950

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 .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions

github-actions Bot commented Nov 7, 2026

Copy link
Copy Markdown
Contributor

First Principles Review (Opus 5.5) — ✅ PASS

Premise-level review of 76d5fbf314f2c0c9f5c7995b31a0a071debe7cc8 — 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

Nothing blocks: each of the ten picks is a fix already on main with a linked user report, and the two pins it rewrites are reversed on those reports.

Notes

  • Item 6 — rides along: docs/build/release.md adds a release-time step (re-pin the two musl sha256s into the EC2 template on every kiro-cli bump), and every future pin bump pays for it. It is derived from the supply-chain pin, and test_kiro_cli_is_the_pinned_release_from_a_host_the_preflight_checks enforces it, so no subtraction is owed.

What this change ships

Inventory (10 items) — 10 justified

Intent: ship ten user-reported fixes that already merged on main in the 0.12.1 patch, one revertible commit each — FIX.

  1. A slow pre-update fetch gets 60 s and answers a retryable 409 instead of a 500 — justified
  2. A mid-wave subagent digest says when its still-running count was taken — justified
  3. Back or Forward onto a closed session shows a "session was closed" notice — justified
  4. Polly and cloud aws children no longer inherit the gateway's PYTHONPATH/PYTHONHOME — justified
  5. Paths with CJK or other non-ASCII punctuation become file chips — justified
  6. The EC2 bootstrap downloads the pinned kiro-cli from prod.download.cli.kiro.dev and checks its SHA-256 — justified
  7. A Node backend's npm deps are installed in every package.json directory down to the entry point — justified
  8. A failed onInstall leaves its redacted output tail in the gateway log — justified
  9. A cancelled side question keeps its text while the cancel is still waiting — justified
  10. The deployment list shows why it cannot run on this host instead of an empty state — justified

[FIRST-PRINCIPLES-REVIEWED] 76d5fbf

Description read

[DESCRIPTION-READ] 1609730dca4661ea72bcab19439bb26c19400821ef358a74c5a91fa0cdee8039

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

github-actions Bot commented Nov 7, 2026

Copy link
Copy Markdown
Contributor

GPT 6.1 Review — ✅ no blocking findings

GPT 6.1 completed its review of 76d5fbf314f2c0c9f5c7995b31a0a071debe7cc8 and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] 76d5fbf

False positive or not applicable? A repository writer can comment:
/ai-review override gpt 76d5fbf314f2c0c9f5c7995b31a0a071debe7cc8 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 7, 2026
@github-actions

github-actions Bot commented Nov 7, 2026

Copy link
Copy Markdown
Contributor

Opus 5.5 Review — ✅ no blocking findings

Reviewed 76d5fbf314f2c0c9f5c7995b31a0a071debe7cc8 — this comment is updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] 76d5fbf

Verdict parsed from the review's SHA-scoped output markers for commit 76d5fbf314f2c0c9f5c7995b31a0a071debe7cc8.

False positive or not applicable? A repository writer can comment:
/ai-review override fable 76d5fbf314f2c0c9f5c7995b31a0a071debe7cc8 false-positive: <reason, 20 to 500 characters>

@buluoray
buluoray merged commit 9c00903 into release/0.12.0 Nov 7, 2026
20 of 24 checks passed
@buluoray
buluoray deleted the crew/slack/0.12.1-batch-20 branch November 7, 2026 03:40
@github-actions github-actions Bot removed the readiness: action required A blocking check or review needs attention label Nov 7, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants