Skip to content

fix(kas): backport the checked-deny grant and kiro-cli toolsSettings grant fixes to 0.12.1 (batch 37) - #32467

Merged
buluoray merged 2 commits into
release/0.12.0from
crew/slack/0.12.1-batch-37
Nov 7, 2026
Merged

buluoray merged 2 commits into
release/0.12.0from
crew/slack/0.12.1-batch-37

Conversation

@buluoray

@buluoray buluoray commented Nov 7, 2026 •

Copy link
Copy Markdown
Contributor

Touches the permission path; needs Bolin's OK before merge. #31622 and #31850 change when Crew answers a KAS built-in tool call from a spec's own grants, and both are hand-ported without #31622's main prerequisites (listed under What changed).

Problem / Motivation

Goal: Bring #31622 and #31850, two KAS permission fixes that merged on main after v0.12.0, onto release/0.12.0 as one commit each in a PR of their own, so they can be reviewed and reverted apart from the other batches.

Why it matters

PR What a 0.12.0 user sees today
#31622 On KAS, an agent whose spec pairs a grant (shell.allowedCommands, or allowedTools naming the shell or write tool) with deniedCommands or a write deniedPaths asks before every call, even one no deny names, because the deny list is read as the author's restriction. An unattended run stalls on the first card. kiro-cli runs those calls unasked and refuses what a deny names.
#31850 On KAS, a kiro-cli agent's shell.allowedCommands beside denyByDefault or autoAllowReadonly asks for every matched command, denyByDefault refuses read-only commands kiro-cli runs, and a write inside write.allowedPaths asks before every write. Reported by several users of unattended agents. kiro-cli runs all of these without asking.

Not a goal

What changed (motivation → approach → change)

Each main squash commit was applied with cherry-pick -n onto release/0.12.0 @ 344cc7ca44, in main order, and committed with its cherry picked from trailer.

PR main commit release commit Pick
#31622 honour a grant beside deniedCommands or deniedPaths 817a9ef363 34bff402fe hand-ported, five files resolved by hand
#31850 keep a kiro-cli agent's allowedCommands, autoAllowReadonly and allowedPaths grants e0e6196e56 1711769359 three files resolved by hand

What it does. When Crew can read every entry of a deniedCommands / write deniedPaths list itself, the projection records the grants that deny alone held back, with the deny list, as one Crew-only value the session stamps on its own agent's requests. A new spec-denies gate row refuses a call a deny names; spec-builtin-grant honours the recorded grant only after the same check answers a clear no. An unreadable entry, a non-plain command, or a target it cannot resolve grants nothing, so the call keeps asking. resolve_managed_entries moves into acp/kas_managed_entries.py for the line cap; kas_agents re-exports it.

What #31850 does. denyByDefault and a boolean autoAllowReadonly beside allowedCommands, and fallbackAction beside a write allowedPaths, no longer withhold that grant (kas_grant_only_settings.without_shell_grant_keys / without_write_grant_keys); deniedCommands, deniedPaths, any other key or a malformed value still do. autoAllowReadonly: true beside denyByDefault lets a read-only command past the spec-shell-allowlist row for the read-only row to answer. A write allowedPaths is recorded and spec-builtin-grant approves a write only when every target is covered as written and symlink-resolved, by the spec-write-allowlist row's own test; a delete is never granted this way. Its pure move (definitions to acp/kas_env_withhold.py and acp/kas_tool_rows.py, re-exported) is in the same squash commit; every moved definition is source-identical on this branch.

Prerequisites checked. Already on this branch: #30961 (batch 32) and #30813 (batch 35). Not taken:

Prerequisite Why not
#30193 (toolsSettings alias spellings, wide path denies, pwsh process control) 16 files; changes how every toolsSettings reader sees alias spellings, and modifies three files this branch does not have (#30108, #30088), so it would pull more in. #31622 needs only its alias folding, in one reader.
#30964 (batch write guard) 720 lines of separate auto-approve behaviour; #31622 only sits next to its CREW_ANSWERED_WRITES_KEY.
#29543 (Crew-owned GitHub OAuth, flag-gated) 3.5k-line feature; only a neighbouring import.

Hand-resolved hunks.

File Resolution
acp/kas_spec_denies.py reads toolsSettings as given instead of folding alias spellings with #30193's canonical_tools_settings. Every spelling's denies are read, and any other key under them still withholds, so it fails toward ask. crew_answered_commands takes one argument on this branch.
acp/kas_agents.py (imports) took the kas_managed_entries import only; this branch keeps _pin_legacy_negotiation in the module
acp/kas_agents.py (spec_keys_without_carrier) added and not _commands_lifted_by_denies(spec) beside release's _allowed_commands_withheld(spec, None)
acp/kas_agents.py (to_client_custom_agent) kept release's tools_settings_to_permissions(spec.get("toolsSettings")) and added untranslated_policy
acp/kas_permissions.py added CREW_SPEC_DENIES_KEY only, to the constants and to the Crew-only key set
docs/agent-spec-fields.md added #31622's one sentence to the toolsSettings row
test/test_kas_spec_denies.py the alias test (needs #30193) is replaced by one that checks a deny under either shell spelling refuses and the rest is granted
#31850: acp/kas_agents.py (imports) dropped the kas_legacy_negotiation import and kept _pin_legacy_negotiation in the module; took CREW_ANSWERED_PATHS_KEY and shell_denies_carried only (CREW_ANSWERED_WRITES_KEY and projection_accepts serve #30964 / #30193 code this branch lacks)
#31850: acp/kas_agents.py (tools_settings) this branch has no tools_settings local in to_client_custom_agent (#30193 adds it), so it binds tools_settings = spec.get("toolsSettings") once and passes it to tools_settings_to_permissions, crew_answered_commands and every _allowed_commands_withheld call, spec_keys_without_carrier included
#31850: acp/kas_permissions.py tool_names import without KIRO_CLI_NAME_BY_KAS_TOOL (moved with the tool rows) and without the alias module; kept this branch's _entry_* readers in kas_agents; aliased _SHELL_SETTINGS_KEYS to SHELL_SETTINGS_KEYS; added crew_answered_paths and kept this branch's crew_answered_builtins signature
#31850: docs/agent-spec-fields.md this branch's toolsSettings row with #31850's own edits applied

The moved resolve_managed_entries matches this branch's copy except its docstring and the constant's name. In each commit, every other file has added and removed lines identical to its main commit. The commit is authored with the public identity the push guard requires and names its original author in its message.

Other open batches. This batch touches acp/harness/kas.py, acp/kas_agents.py and docs/agent-spec-fields.md, which #32229 also touches, and docs/agent-spec-fields.md, which #32349 also touches. git merge-tree of this batch on top of each of their heads reports no conflict.

builder-security checked: no bind, listener, route, auth mechanism or storage grant changes. The changes widen auto-approve cases on KAS for an agent's own spec grants, bounded as described above, and keep the deny floor, the deny rows and governance ahead of them.

Backwards compatibility

Compatible: the changes are the ones already merged on main, and their own compatibility notes apply unchanged. An agent with a grant beside a deny list Crew can read now runs what no deny names without asking, as on kiro-cli; a deny Crew cannot read keeps the old prompt. A read-only command under denyByDefault with autoAllowReadonly: true now runs, as kiro-cli runs it.

Two public functions move out of their modules for the line cap; each old module re-exports its name, so every reader keeps its import path:

Reader: src/kiro_crew/acp/harness/kas.py:resolve_managed_entries -- chat -- reads kas_agents.resolve_managed_entries, re-exported from acp/kas_managed_entries.py; test_kas_agents_mcp_servers.py
Reader: src/kiro_crew/agent_sdk/spec_hooks.py:kas_tool_match_names -- chat -- imports it from kas_permissions, which re-exports it from acp/kas_tool_rows.py; test_spec_hook_kas_tool_names.py, test_harness_tool_names.py

Tests

Proof: not provable -- other: a backport of a main commit whose proof ran on its own PR

  • 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 names a changed module or record (kas_spec_denies, spec_denied, kas_managed_entries, kas_grant_only_settings, kas_spec_grants, kas_env_withhold, kas_tool_rows, kas_write_allowlist, allowed_commands, allowed_paths, gate_tiers, tool_identity, acp_turns, the KAS session handle modules, spec_granted_paths, spec_shell_readonly or to_client_custom_agent): 139 files, one pytest process per batch of 20, -n 0. 8661 passed. 21 failed, and the same 21 fail on a clean release/0.12.0 checkout with the same command (copilot backend, harness-parity is_kiro_cli, doctor, and the conductor verify-fix contract suite).
  • The two source PRs' own test files (test_kas_spec_denies.py, test_kas_tools_settings_grants.py, test_kas_allowed_commands.py, test_kas_write_fallback_deny.py, test_kas_tools_settings.py, test_hooks.py, test_crew_fired_spec_hooks.py): 459 passed.
  • Dropping fallbackAction from the write grant-only keys turns 2 of those tests red on this branch.
  • flake8 passes on the changed Python files.

Manual verification

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

Screenshots / video

Related Issues

Backport to the 0.12 line for 0.12.1 of #31622.

Sibling sweep

Siblings: none -- the source PR carried its own sibling sweep; this PR only carries that fix.

Pattern harvest

Not generalizable: this PR only carries a fix already on main. The 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.

A KAS spec that pairs allowedCommands (or an allowedTools shell or
write grant) with deniedCommands or a write deniedPaths asked for every
call, because Crew read the deny as the author's restriction. On
kiro-cli the grant runs unless a deny names the call.

When Crew can read every deny entry itself, the projection records the
grants the deny alone held back together with the deny list, as one
value the session stamps on its own requests. A new spec-denies gate
row refuses a call a deny names; spec-builtin-grant honours the grant
only after the same check answers a clear no. Anything unreadable
keeps asking.

(cherry picked from commit 817a9ef)

Release backport, hand-ported without its main prerequisites (#30193
toolsSettings aliases, #30964 batch write guard, #29543 GitHub OAuth),
which are wide and change other behaviour:
- acp/kas_spec_denies.py: reads toolsSettings as given instead of
  folding kiro-cli alias spellings (#30193). Every spelling's denies are
  read and any other key under them still withholds, so this fails
  toward ask; crew_answered_commands takes one argument here.
- acp/kas_agents.py: took the kas_managed_entries import only (this
  branch keeps _pin_legacy_negotiation in the module);
  spec_keys_without_carrier adds the lift beside release's
  _allowed_commands_withheld(spec, None); to_client_custom_agent keeps
  release's tools_settings_to_permissions(spec.get("toolsSettings")) and
  adds untranslated_policy.
- acp/kas_permissions.py: added CREW_SPEC_DENIES_KEY only (the
  CREW_ANSWERED_WRITES_KEY beside it is #30964's).
- docs/agent-spec-fields.md: added this change's one sentence to the
  toolsSettings row.
- test/test_kas_spec_denies.py: the alias test is replaced by one that
  checks a deny under either shell spelling refuses.
Every other file applied unchanged.

Original author: Ray Xu (#31622).
…and allowedPaths grants (#31850)

* refactor(kas): move env-withholding and tool-row helpers to sibling modules

kas_agents.py and kas_permissions.py sit just under the 2,000-line cap,
so the KAS toolsSettings grant fix that follows cannot land in them.
This commit only moves whole definitions, with no behaviour change:

- kas_agents -> kas_env_withhold: _MANAGED_ENV_KEYS_KEPT,
  _ENV_REFERENCE, _withhold_credential_fields, _is_relayed_reference,
  _carries_env_reference.
- kas_permissions -> kas_tool_rows: _HOOK_ONLY_ROWS,
  _rows_from_policy_table, KAS_TOOL_IDS_BY_KIRO_TOOL,
  KAS_TOOL_MATCH_VOCABULARY, kas_tool_match_names.

Both old modules re-export every name they held, so callers and test
patch points keep their import paths. Each moved definition is
source-identical to its original. The managed-server helpers
(resolve_managed_entries and its neighbours) are left in place: another
change moves them.

* fix(kas): keep a kiro-cli agent's allowedCommands, autoAllowReadonly and allowedPaths grants

(cherry picked from commit e0e6196)

Release backport, on top of #31622's hand-port. The definitions this
change moves to acp/kas_env_withhold.py and acp/kas_tool_rows.py are
source-identical on this branch. Hand-resolved:
- acp/kas_agents.py: dropped the kas_legacy_negotiation import and kept
  _pin_legacy_negotiation in the module (this branch never moved it);
  took CREW_ANSWERED_PATHS_KEY and shell_denies_carried only (the
  CREW_ANSWERED_WRITES_KEY and projection_accepts imports serve #30964
  and #30193 code this branch does not have). This branch has no
  tools_settings local in to_client_custom_agent (#30193 adds one), so
  it binds tools_settings = spec.get("toolsSettings") once and passes it
  to tools_settings_to_permissions, crew_answered_commands and every
  _allowed_commands_withheld call, including spec_keys_without_carrier.
- acp/kas_permissions.py: tool_names import without
  KIRO_CLI_NAME_BY_KAS_TOOL (it moved with the tool rows) and without
  the tools_settings_aliases module; kept this branch's _entry_* readers
  in kas_agents and aliased _SHELL_SETTINGS_KEYS to SHELL_SETTINGS_KEYS;
  added crew_answered_paths and kept this branch's
  crew_answered_builtins signature.
- docs/agent-spec-fields.md: this branch's toolsSettings row with this
  change's own edits applied.
Every other file applied unchanged.

Original author: Ray Xu (#31850).
@buluoray
buluoray requested review from a team and KiroCrewCaptain as code owners November 7, 2026 16:47
@buluoray
buluoray requested review from tim-finnigan and removed request for a team November 7, 2026 16:47
@github-actions

github-actions Bot commented Nov 7, 2026

Copy link
Copy Markdown
Contributor

Sensitive Change Review

Sensitive files changed:

  • src/kiro_crew/hook_runtime/gate_tiers.py

  • src/kiro_crew/hook_runtime/tool_identity.py

  • src/kiro_crew/hooks.py

  • No reasoning template for 1711769 yet.

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: 17117693599f21b20504fd6d25da2f3f4f8cd549
- 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 7, 2026
@github-actions

github-actions Bot commented Nov 7, 2026

Copy link
Copy Markdown
Contributor

UX Review (Opus 5.5) — ⏭️ skipped

Revision 17117693599f21b20504fd6d25da2f3f4f8cd549 touches no user-facing surface (no changes under website/ or committed screenshots), so the UX review was skipped. Does not block merge.

@github-actions

github-actions Bot commented Nov 7, 2026

Copy link
Copy Markdown
Contributor

Design Review (Opus 5.5) — ✅ PASS

Design-level review of 17117693599f21b20504fd6d25da2f3f4f8cd549 — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS does not.

Design-Verdict: PASS

Nothing blocks: this backports two KAS grant fixes already merged on main into one commit each. The tests it rewrites pinned the old "ask" behaviour, and main already changed those same tests.

Notes

[DESIGN-REVIEWED] 1711769

Description read

[DESCRIPTION-READ] 05d5d4ee42591d8755a785ad437ab55abe151a9b622f1a25a825241d6693631c

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 17117693599f21b20504fd6d25da2f3f4f8cd549 — 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. This backports two KAS permission fixes already merged on main. Every new carrier has a gate-row consumer, and the moved definitions are line-cap splits that keep their re-exports.

What this change ships

Inventory (8 items) — 8 justified

Intent: let a KAS agent's own spec grants run calls unasked the way kiro-cli runs them, so unattended runs on 0.12.x stop stalling on approval cards. This is a FIX (a backport of #31622 and #31850).

  1. On KAS, a shell or write grant next to a deniedCommands / write deniedPaths list that Crew can fully read no longer asks before every call — justified
  2. A new spec-denies gate row refuses any call that a spec deny names — justified
  3. allowedCommands next to denyByDefault, or next to a boolean autoAllowReadonly, now runs matched commands unasked — justified
  4. With autoAllowReadonly: true next to denyByDefault, read-only commands get past the shell allowlist row — justified
  5. A write inside every write.allowedPaths target is answered at the gate, never a delete — justified
  6. The session-start notice stops naming allowedPaths and boolean autoAllowReadonly once they are carried — justified
  7. Three new Crew-only projection keys and their AcpEvent fields; each counted 6–9 consumer files under src/ — justified
  8. resolve_managed_entries, the env-withhold definitions and the tool rows move into sibling modules and are re-exported — justified

[FIRST-PRINCIPLES-REVIEWED] 1711769

Description read

[DESCRIPTION-READ] 742f44401f4b66066bbd5e7b8e4eed233bceee241475c0314dd3e68fc6b1cdb9

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

Security Scope Review -- ✅ PASS

Which legitimate operations does 17117693599f21b20504fd6d25da2f3f4f8cd549 newly refuse? Updated in place on each push. A script-confirmed regression blocks; the model proposes candidates and never decides a refusal.

Verdict basis: adjudicated, zero confirmed regressions.

Security scope review -- adjudicated candidates

  • macos-latest (platform posix): 11 classified, 0 newly refused, 11 still allowed.
  • ubuntu-latest (platform posix): 11 classified, 0 newly refused, 11 still allowed.
  • windows-latest (platform windows): 5 classified, 0 newly refused, 5 still allowed.

No candidate flipped from allowed to refused.

Scope-Verdict: PASS

Reason: The PR leaves the three shell deny tiers (sensitive-bash, exfil, deny-rules) as they are, so I expect the candidates to confirm no newly refused commands. The one new refusal is the spec-denies gate row. It only fires on a KAS request that carries a SpecDeniedGrant, and the classifier can't build that context, so this row is named below as unadjudicated rather than counted as a pass. It reads a deny wider than the anchored glob that used to travel, so some calls that used to ask now get refused outright.

TIGHTENING: a new deny_policy gate row, GateTier("spec-denies", "deny_policy", _tier_spec_denies), backed by spec_denies_refusal -- it refuses a KAS agent's own shell call that a spec deniedCommands entry names "whole or in a leading piece that ends at a word" (denied_match(..., word_bounded=True)). It also refuses a write whose target, "in ANY of its spellings (as written, normalized, or symlink-resolved)", a write deniedPaths entry covers (_path_denied). The threat it refuses: once the PR stops sending such a spec to the prompt, a lifted grant could run a command or write that the spec's author denied.

ADJUDICATED: 16 candidates, 0 confirmed newly-refused

PLATFORM COVERAGE: I wrote posix, windows and no any rows. The posix rows probe the shell surface the PR's own tests use: make build, the make cleanup / make clean-all mid-word neighbours, npm test / npm publish --dry-run, the reported kubectl get and ./scripts/report.sh grants, the autoAllowReadonly reads grep/cat, rm, and git log. The windows rows cover the spellings the PR assumes about separators and executable names: .\scripts\report.cmd, kubectl.exe, type, findstr with \ paths, and del. The PR's write-path assumptions are Windows drive and \ forms read through allowed_paths.to_glob_path, plus symlink resolution. No shell row can express them, so they are listed under UNADJUDICATED.

UNADJUDICATED:

  • Guard: the spec-denies row (hooks.GATE_TIERS, after spec-write-allowlist).
    • Call path: KAS projection to_client_custom_agent → spec_denies_record writes _crewSpecDenies → KasHarness.record_session_projection / switched_spec_denies sets handle.kas_spec_denies → _build_permission_event stamps AcpEvent.spec_denied_grant → ToolCall.from_event → _tier_spec_denies → spec_denies_refusal.
    • It is not in SHELL_DENY_TIERS, so no shell row can reach it.
  • Possible narrowing I can argue but not confirm:
    • kiro-cli anchors a deniedCommands regex \A..\z; base _regex_to_glob says so, and the travelling KAS glob matches it whole. The new refusal also matches every word-ending leading piece.
    • So with deniedCommands ["make clean"], a call such as make clean now, or npm publish under a deny of npm, used to reach the prompt. When the row applies, it now hard-fails on every approval policy, including scheduled approval_mode: auto jobs.
    • The PR's docstring says kiro-cli reads such an entry "as a prefix", which contradicts the base \A..\z comment. Which reading kiro-cli actually uses is the open question.
    • Inputs I would have submitted, each with that spec context:
      • make clean now and make clean build under ["make clean"]
      • npm publish --dry-run under ["npm publish"]
      • kubectl get secrets-report under ["kubectl get secrets"]
  • Write half: _path_denied refuses when any spelling, or any entry's link-resolved twin, is covered. That is wider than how an allowlist is read.
    • Inputs I would have submitted, each with a deniedPaths entry on the session:
      • a write to out/report.md, where out is a symlink into a denied directory the author meant only for its real path
      • a Windows write to C:\work\Out\a.txt against a deny of c:/work/out/secret/**, to check the case-folding boundary
      • a relative target with no session directory, which returns unknown and only withholds the grant

[SCOPE-REVIEWED] 1711769

@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 17117693599f21b20504fd6d25da2f3f4f8cd549 and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] 1711769

False positive or not applicable? A repository writer can comment:
/ai-review override gpt 17117693599f21b20504fd6d25da2f3f4f8cd549 false-positive: <reason, 20 to 500 characters>

@github-actions

github-actions Bot commented Nov 7, 2026

Copy link
Copy Markdown
Contributor

Opus 5.5 Review — ✅ no blocking findings

Reviewed 17117693599f21b20504fd6d25da2f3f4f8cd549 — this comment is updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] 1711769

Verdict parsed from the review's SHA-scoped output markers for commit 17117693599f21b20504fd6d25da2f3f4f8cd549.

False positive or not applicable? A repository writer can comment:
/ai-review override fable 17117693599f21b20504fd6d25da2f3f4f8cd549 false-positive: <reason, 20 to 500 characters>

@github-actions github-actions Bot added readiness: passed Eligible automated validation passed for the current revision and removed readiness: checking Automated validation is still running labels Nov 7, 2026
@buluoray
buluoray merged commit 37ecd16 into release/0.12.0 Nov 7, 2026
33 of 39 checks passed
@buluoray
buluoray deleted the crew/slack/0.12.1-batch-37 branch November 7, 2026 17:10
@github-actions github-actions Bot removed the readiness: passed Eligible automated validation passed for the current revision 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.

1 participant