Skip to content

chore: 0.13 backport, bump the bundled KAS engine to 0.66.117 on Node 22.23.4 (#38228, #39211) - #39376

Merged
bolichen97 merged 1 commit into
release/0.13.0from
fix/0.13-backport-kas-node-22.23.4
Nov 19, 2026
Merged

bolichen97 merged 1 commit into
release/0.13.0from
fix/0.13-backport-kas-node-22.23.4

Conversation

@bolichen97

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: The 0.13.0 release installs the same KAS engine as main: 0.66.117 on Node 22.23.4.

release/0.13.0 pins KAS 0.66.104, which ships Node 22.23.3. Main moved to 0.66.117 in #38228 and to the Node 22.23.4 republish of it in #39211. Node 22.23.4 is the 2026-10-21 security release for the 22.x line.

Why it matters

The bundled Node is a private runtime users cannot patch themselves: the manifest pins every Node build by sha256. Without this backport, 0.13.0 users keep Node 22.23.3 for a full release cycle.

Not a goal

  • Publishing anything: both bundles are already on the CDN.
  • Changing the engine build beyond what main ships.
  • Cutting a new release candidate or tag (the release owner does that after merge).

What changed (motivation → approach → change)

Pick Main PR Effect
1 #38228 Pins KAS 0.66.117. The engine now sends the model select after session/new, so the start path reads it from the notification stream and settles it before the first prompt. Bundle fixtures re-captured on 0.66.117.
2 #39211 Pins 0.66.117-node22.23.4, the same engine build republished with Node 22.23.4. KAS_PIN_ENGINE_RELEASE strips the -node suffix wherever engine versions are compared.

Both are git cherry-pick -x picks squashed into one commit, with no conflicts and no edits. #39211 needs #38228 first, because its pin is a republish of 0.66.117. The picks touch acp/kas_pin.py, acp/kas_bundle.py, acp/kas_engine_features.py, the acp/session_handle_* model-select and MCP files, acp/runtime_session_load.py, agent_sdk/kas_model_catalog.py, cli_parser_gateway.py (the kas install --version help text), the bundle fixtures and their tests, and the two specs, same as on main.

Backwards compatibility

Compatible: the installer resolves whatever directory the pin names, and kas/0.66.104/ stays published, so older builds keep installing.

Removes nothing.

Tests

  • The picks bring their own tests (test/test_kas_pin.py, test/test_kas_model_catalog.py, test/test_kas_model_select_capture.py, test/test_harness_parity.py and the other updated KAS and session tests), same as chore: bump the bundled KAS engine pin to 0.66.117 #38228 and chore: bump the bundled kas node runtime to 22.23.4 #39211.
  • Local on this head: isort, flake8 and mypy pass. The changed tests plus every test_kas_*, test_acp_frame*, test_session_handle* and test_real_adapter* file: 6317 passed, 33 skipped, 1 failed. The one failure (test_kas_cli_floor.py::TestTheProbeRunsTheSpawnBinary::test_a_timed_out_probe_kills_the_whole_probe_tree, os.pidfd_open missing) fails the same way on clean release/0.13.0: the local Python build has no pidfd_open.

Manual verification

N/A: the full CI matrix on this release branch is the check a backport needs.

Screenshots / video

N/A: no UI change.

Related Issues

Part of #38226

Part of #20727

Backport of #38228 and #39211 to release/0.13.0.

Sibling sweep

Swept: later main commits to src/kiro_crew/acp/kas_pin.py
Sibling: none -- #39211 is the latest change to the pin on main

Pattern harvest

Not generalizable: an engine pin backport; both picks apply cleanly.

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

Contributed under the project's Apache-2.0 license.

… 22.23.4 (#38228, #39211)

Backports two main PRs to release/0.13.0, in pick order:

- #38228 chore: bump the bundled KAS engine pin to 0.66.117
  (cherry picked from commit b490cdf)
- #39211 chore: bump the bundled kas node runtime to 22.23.4
  (cherry picked from commit 5525caa)

Both picks applied without conflicts and without edits. #39211 pins the
0.66.117 engine republished with Node 22.23.4, so it needs #38228 first.

Nothing left out.
@bolichen97
bolichen97 requested review from a team and KiroCrewCaptain as code owners November 19, 2026 03:23
@bolichen97
bolichen97 requested review from chenmingwei23 and removed request for a team November 19, 2026 03:24
@github-actions github-actions Bot added the readiness: checking Automated validation is still running label Nov 19, 2026
@github-actions

Copy link
Copy Markdown
Contributor

UX Review (Opus 5.5) — ⏭️ skipped

Revision 81bb3c3f430207eb002be01ffcc0c635d3320df2 touches no user-facing surface (no changes under website/ or committed screenshots), so the UX review was skipped. Does not block merge.

@github-actions

Copy link
Copy Markdown
Contributor

First Principles Review (Opus 5.5) — ✅ PASS

Premise-level review of 81bb3c3f430207eb002be01ffcc0c635d3320df2 — 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 is a straight backport of two main PRs. Every item comes either from the measured engine change or from the Goal of matching main, so nothing blocks.

Notes

  • Item 4 — zero consumers: the description says KAS_PIN_ENGINE_RELEASE "strips the -node suffix wherever engine versions are compared". A grep for KAS_PIN_ENGINE_RELEASE in src/ finds 0 consumers outside its own definition and __all__; only test/test_real_adapter_fixture_pins.py and test/test_kas_pin.py read it. Subtraction: none here, since the Goal is to match main exactly. Delete it on main, or make it private to the tests.

What this change ships

Inventory (6 items) — 5 justified

Intent: 0.13.0 users get the same patched KAS engine and Node runtime as main. This is a FIX (dependency/security bump), backported.

  1. 0.13.0 installs KAS 0.66.117 with bundled Node 22.23.4 instead of 0.66.104 with Node 22.23.3 — justified
  2. A new KAS session waits up to 2 s for the late model select and runs the served-default check before the first prompt — justified
  3. A config_option_update on KAS now refreshes the model list as well as effort levels — justified
  4. A new exported constant KAS_PIN_ENGINE_RELEASE (the pin without its -node suffix) — zero consumers (tests only)
  5. The kas install --version help example and the docstrings name 0.66.117 — justified
  6. The bundle fixtures were re-captured on 0.66.117, and the two specs were updated — justified

[FIRST-PRINCIPLES-REVIEWED] 81bb3c3

Description read

[DESCRIPTION-READ] 82bdd1ddca5c9c80762846eb138d88eaf95b0f5d5d1b709fa6575ec33504bdb9

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

Design Review (Opus 5.5) — ✅ PASS

Design-level review of 81bb3c3f430207eb002be01ffcc0c635d3320df2 — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS does not.

Design-Verdict: PASS

Nothing blocks: this backports #38228 and #39211 onto the release branch exactly as they landed on main, and the -node suffix pin is handled where engine versions are compared.

Notes

  • The description says KAS_PIN_ENGINE_RELEASE "strips the -node suffix wherever engine versions are compared." In src/, though, nothing reads it outside kas_pin.py; only tests use it. engine_sends_model_select_late(bundle.version) still works because parse_kiro_cli_version's _VERSION_RE.match reads only the leading part of the version string, which happens to drop the suffix. Using the constant (or a stated parse rule) at the comparison sites would make the claim true by design, but that change belongs on main first.

[DESIGN-REVIEWED] 81bb3c3

Description read

[DESCRIPTION-READ] 45fdc3e15a3da681839129af3212dfbfa35ceb89b4d4868a6f6caa24a18d9630

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

Copy link
Copy Markdown
Contributor

Opus 5.5 Review — ✅ no blocking findings

Reviewed 81bb3c3f430207eb002be01ffcc0c635d3320df2 — this comment is updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] 81bb3c3

Verdict parsed from the review's SHA-scoped output markers for commit 81bb3c3f430207eb002be01ffcc0c635d3320df2.

False positive or not applicable? A repository writer can comment:
/ai-review override fable 81bb3c3f430207eb002be01ffcc0c635d3320df2 false-positive: <reason, 20 to 500 characters>

@github-actions

Copy link
Copy Markdown
Contributor

GPT 6.1 Review — ✅ no blocking findings

GPT 6.1 completed its review of 81bb3c3f430207eb002be01ffcc0c635d3320df2 and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] 81bb3c3

False positive or not applicable? A repository writer can comment:
/ai-review override gpt 81bb3c3f430207eb002be01ffcc0c635d3320df2 false-positive: <reason, 20 to 500 characters>

@bolichen97
bolichen97 merged commit 7f98f1d into release/0.13.0 Nov 19, 2026
88 of 89 checks passed
@bolichen97
bolichen97 deleted the fix/0.13-backport-kas-node-22.23.4 branch November 19, 2026 04:05
@github-actions github-actions Bot removed the readiness: checking Automated validation is still running label Nov 19, 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.

1 participant