Skip to content

chore(release): backport the 0.11.1 hotfix fixes to release/0.11.0 - #26487

Merged
bolichen97 merged 1 commit into
release/0.11.0from
backport/0.11.1-hotfix
Oct 31, 2026
Merged

bolichen97 merged 1 commit into
release/0.11.0from
backport/0.11.1-hotfix

Conversation

@bolichen97

@bolichen97 bolichen97 commented Oct 31, 2026 •

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: Ship the P0 and must-have fixes already merged on main to the 0.11.0 release line, so the 0.11.1 hot patch can be cut from it.

These main PRs merged after release/0.11.0 was cut, so no 0.11.0 build has them. Users on 0.11.0 still hit a split or corrupt memory.db, a stop the desktop app undoes, stuck approvals, blank KAS resumes and the other bugs below.

Why it matters

The release line is what users install. Each fix here is already reviewed and merged on main; without this PR it reaches users only with 0.12.0.

Not a goal

  • No change beyond what each main PR ships; no new behaviour.
  • No version bump, CHANGELOG entry, RC or tag: the 0.11.1 release PR and a person do those.
  • Not every main PR since the cut: only the P0 list below.

What changed (motivation → approach → change)

One commit: each main PR below was git cherry-pick -x'd in this order, then squashed (PR Hygiene allows two commits). The commit message lists every pick and its source sha.

  1. fix(memory): read memory.db off the writer lock, split background pool #24290: read memory.db off the writer lock (executors.py, vector_memory.py, vector_memory_lifecycle.py, history_consolidation_pass.py, slack/gateway_runtime/memory_lifecycle.py).
  2. fix(dashboard): bound the pre-prompt build's wait to enter the memory pool #24414: bound the pre-prompt wait for the memory pool (dashboard/chat_turn/prompt_assembly.py).
  3. fix(memory): keep the shared -wal when another process closes the memory store #25852: keep the shared -wal on close (sqlite_wal.py, diagnose_probes.py, one .github/black-baseline.txt row).
  4. fix: read the page-cache prefetch in a child to keep sqlite locks #26302: page-cache prefetch in a child (page_cache.py).
  5. perf(memory): score facts from a term index and prebuild after restart #25328: term-index scoring, prebuild after restart (vector_memory_runtime/term_index.py, memory_startup_prebuild.py).
  6. fix(desktop): retry a failed netstat before Windows recovery gives up #25792: retry netstat in Windows recovery (website/electron).
  7. fix(cli): restart says a gateway still serves the port when refused #25366 and 8. fix(cli): refuse to stop or restart a gateway the desktop app supervises #25705: restart/stop and a desktop-supervised gateway (cli_server.py, cli_server_restart.py, cli_server_stop.py, apps/builtins/dev_fleet).
  8. fix(hooks): approve kiro-cli read-only file searches after safety checks #25150: approve kiro-cli read-only searches (hook_runtime/gate_tiers.py, hooks.py).
  9. fix(settings): refuse a config save that config.local.json overrides #26216: refuse a save config.local.json overrides (dashboard/handlers, website/src/pages/settings, website/src/test/ChatPanel.defaultModel.test.tsx).
  10. fix(workflows): write non-finite floats as null so runs open #26085: non-finite floats as null (workflows/store.py).
  11. fix(approvals): give each background tool approval its own id #25905: one id per background approval (slack/gateway_runtime/turn_approvals.py).
  12. fix(chat): record file changes for kiro-cli v3 write tools #25650: file changes for kiro-cli v3 writes (dashboard/chat_turn/file_changes.py).
  13. fix(sandbox): stop blaming a host-wide OOM kill on a per-tree limit #25647: OOM blame (sandbox_slice.py).
  14. fix(acp): replay history when a KAS resume restores no turns #25521: KAS resume replays history (acp/runtime_session_load.py, providers/acp_runtime_start.py).
  15. fix(kas): auto-approve granted built-ins after Crew's own checks #25339: KAS auto-approves granted built-ins after Crew's checks (acp/kas_permissions.py, platform/governance.py).
  16. fix(kas): mount web_search as the engine's remote_web_search #26435: KAS mounts web_search as remote_web_search (platform/tool_names.py).
  17. fix(kas): answer a spec's shell.allowedCommands at the gate, after Crew's checks #25539: KAS answers shell.allowedCommands at the gate (platform/allowed_commands.py).
  18. fix(kas): relay an MCP server's boolean env flags, never its secrets #25331: KAS relays boolean MCP env flags, never secrets (acp/kas_agents.py, security_posture.py).
  19. fix(security): read the -c script past options and redirections #26087: the shell reader finds the -c script past options (security/shell_normalizer.py).
  20. fix(kas): refuse a shell command a denyByDefault spec does not allow #26501: KAS refuses a shell command a denyByDefault spec does not allow (hook_runtime/tool_identity.py).
  21. fix(kas): name a deniedCommands deny aimed at a redirect as not carried #26489: KAS names a deniedCommands deny aimed at a redirect as not carried.
  22. fix(security): delete the curl-upload deny rule #26469: fix(security): delete the curl-upload deny rule.

Adapted by hand (also in the commit message): #26302 and #25328 and #25705 test pins; #26085 one import; #25339 keeps 0.11.0's missing-ceiling rule (main's enterprise clause is absent here); #26435 edits the H6 row on the release text; #25539 drops two doc lines whose paragraphs 0.11.0 lacks; #25331's test uses 0.11.0's hoist_managed_servers(resume=); #26087's security line budget is re-measured at 32,714. #26501 keeps 0.11.0's rule shapes in a comment and its doc line moves into the toolsSettings row; #26489's sentence joins that row.

Left out:

Docs move with their picks (docs/system-specs, docs/architecture, src/kiro_crew/docs, src/kiro_crew/builtin_skills). The test-only fixes are under Tests.

Backwards compatibility

Compatible: each change is its main PR's own, already reviewed for compatibility there; the hand-resolved conflicts only adapt test pins to the release branch.

Removes nothing: no change here takes anything away beyond what the main PRs did, and each was reviewed on main.

Tests

Test fixes

Test fixes (test files only; no assertion is loosened). Each test failed on this host, and also on a clean release/0.11.0 and on main:

  • TMPDIR under ~/.kiro/crew (an agent session's scratch dir) made the code under test treat every test path as the data home. test/tmpdir_helpers.py gains short_tmp_path, used as a tmp_path override in the file-explorer, design-tweak, spawn-offload, launch-copy, home-pin and native-skill tests, and for one auto-improvement-members test; the sage ancestor-chain test (src/kiro_crew/apps/builtins/code_review_sage/tests/test_store_ancestor_chain.py) makes its dir under /tmp.
  • A mise shim under a scratch HOME hangs (node) or writes to stdout (python3.9): the pi-gate and remote-session tests take the first real binary on PATH, as the installer tests already do.
  • The snapshot-merge ACL test names its own uid; nobody cannot be named inside a user namespace.
  • The installer-step and push_guard interrupt tests spawn their child with SIGINT at its default; a background run starts with it ignored.
  • The 30 electron reds were a missing npm ci in website/electron in my worktree, not a test defect; CI installs it. With it, 2874 pass.

The picked tests are the main PRs' own. Run on this tree:

  • isort, flake8, mypy src/kiro_crew/, semgrep (repo rules, diff-scoped), flake-pattern ratchet, file line cap, comment history, check_black_formatting.py, docs-lint.sh and harness parity: all pass.
  • npm --prefix website run test:electron (with website/electron deps installed as CI does): 2874 tests, 0 failed.
  • Every file that failed in the related-set run, plus the files the fixes and the security pick touch: 1399 passed, 0 failed.
  • The changed test files with test_vector_memory_lock_discipline.py, test_spawn_audit.py, test_hooks_composition_contract.py: 2712 passed. test_security*, test_denied*, test_exfil*, test_blocked*: 7256 passed.
  • The interrupt tests also pass when pytest starts with SIGINT ignored.
  • Under host load near 35 on 32 cores, a few path-check tests can miss the 2 s resolver budget; this happened once in 12 loaded runs, and never on a quiet host. No production timeout was changed.
  • Local GPT and Opus review on this head: PASS, no findings (and PASS on each earlier head).

Determinism: the tests are the main PRs' own, unchanged; their repeat and shuffled-order runs are recorded on each main PR.

Manual verification

N/A - every change was verified on its own main PR; this PR moves them to the release branch.

Screenshots / video

Why no screenshot: every frontend change here (#25792, #26216) is a byte-identical cherry-pick of a main PR that carries its own evidence.

Related Issues

Part of #25846

Backport of #24290, #24414, #25852, #26302, #25328, #25792, #25366, #25705, #25150, #26216, #26085, #25905, #25650, #25647, #25521, #25339, #26435, #25539, #25331, #26087, #26501, #26489, #26469. Replaces #26474 and #26465.

Sibling sweep

Siblings: none -- a backport of main fixes; each sibling sweep is on its main PR

Pattern harvest

Not generalizable: backport of main fixes; each harvest is recorded on its main PR.

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.

@github-actions

github-actions Bot commented Oct 31, 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

  • src/kiro_crew/sandbox_slice.py

  • src/kiro_crew/security/denied_rules_catalog.py

  • src/kiro_crew/security/exfil_bash.py

  • src/kiro_crew/security/shell_normalizer.py

  • src/kiro_crew/security_posture.py

  • @KiroCrewCaptain: approved an older commit (2049d09); re-approve on 0ff5fce; 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: 0ff5fcee27154e124a70221f6fd31decb75a2aa2
- 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 Oct 31, 2026
@github-actions

github-actions Bot commented Oct 31, 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 0ff5fcee27154e124a70221f6fd31decb75a2aa2 — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS does not.

UX-Verdict: PASS

cannot evaluate: missing a screenshot of the new settings-row error shown when config.local.json overrides a setting. No blind read ran because the PR has no images.

Evidence gaps

  • New error state on the Settings → chat panel model and effort rows (useChatPanelModels.ts withOverlayReason, which shows the server's 409 config_overlay_owned text, "<path_key> is set in config.local.json, which overrides this setting. Change or remove it there."). No screenshot shows it, and the blind-read report says "BLIND READ NOT PERFORMED". To close the gap, attach a screenshot of a row after a refused save, with the key set in config.local.json, so we can see the full wording, any truncation, and where it sits next to the picker.

Notes

  • The message starts with a raw config key (agent.default_model is set in…). Lead with the setting's visible label, and keep the key only as the reference to look up in the file.

[UX-REVIEWED] 0ff5fce

Description read

[DESCRIPTION-READ] 9a08b4d2e9d8adf250ceb30fc5c2a0a998fdab4204df32ccb178696ce75d81e0

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 Oct 31, 2026 •

Copy link
Copy Markdown
Contributor

First Principles Review (Opus 5.5) — ✅ PASS

Premise-level review of 0ff5fcee27154e124a70221f6fd31decb75a2aa2 — 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

Every item is a cherry-pick of a fix already merged on main. The only part that isn't is a set of test-only riders, which leaves this branch out of step with main.

Notes

  • Item 10 — rides along, undeclared against the Goal: ## Not a goal says "No change beyond what each main PR ships", but the description says the test fixes failed "also on a clean release/0.11.0 and on main". So short_tmp_path in test/tmpdir_helpers.py and its override fixtures, the mise-shim PATH picks, the uid and SIGINT pins exist on no main PR, and they "fail in agent-driven runs while passing in CI". Subtraction: move these test-file changes out of this backport into their own PR to main, then pick that PR here like the others.

What this change ships

Inventory (10 items) — 9 justified

Intent: ship 23 fixes already merged on main to release/0.11.0 so 0.11.1 can be cut from it. FIX.

  1. memory.db reads move off the writer lock; the pre-prompt pool wait is bounded; the shared -wal survives close — justified
  2. Page-cache prefetch runs in a child; the term index is prebuilt after a restart — justified
  3. restart/stop no longer get undone by a desktop-supervised gateway; Windows recovery retries netstat — justified
  4. kiro-cli read-only searches are approved; a save that config.local.json overrides is refused — justified
  5. Non-finite workflow floats are stored as null; each background approval gets its own id; kiro-cli v3 writes show file changes; OOM blame is fixed — justified
  6. A KAS resume replays history; KAS auto-approves granted built-ins; KAS mounts remote_web_search — justified
  7. KAS answers allowedCommands and denyByDefault at the gate, relays boolean MCP env flags only, and says when a redirect deny is not carried — justified
  8. The shell reader finds the -c script past options — justified
  9. The curl-upload deny rule is gone (data-exfil-curl-upload, from main fix(security): delete the curl-upload deny rule #26469) — justified
  10. Host-dependent test fixes that exist on no main PR — rides along, undeclared against "No change beyond what each main PR ships"

[FIRST-PRINCIPLES-REVIEWED] 0ff5fce

Description read

[DESCRIPTION-READ] bff1c1ca555f45198efddd3c7e19c2ca5a303989c2616614b19341f72073cdd7

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 Oct 31, 2026 •

Copy link
Copy Markdown
Contributor

Design Review (Opus 5.5) — ✅ PASS

Design-level review of 0ff5fcee27154e124a70221f6fd31decb75a2aa2 — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS does not.

Design-Verdict: PASS

Nothing blocks: these are cherry-picks of fixes already merged on main, and the hand-adapted conflicts only touch test pins, an import and doc rows.

Notes

  • ## Backwards compatibility says "Removes nothing", but fix(security): delete the curl-upload deny rule #26469 deletes the data-exfil-curl-upload deny rule (denied_rules_catalog.py, exfil_bash.py, golden fixture). The only remaining mention in the tree is a comment in test/test_denied_commands_security.py:48, so no reader breaks. The section should still list this loosening as a take-away instead of claiming nothing is removed.
  • "Manual verification: N/A" covers fix(desktop): retry a failed netstat before Windows recovery gives up #25792's Windows-only netstat retry. That relies on the main PR's Windows check carrying over, which holds only because the electron files are byte-identical. Say so explicitly for that pick.

[DESIGN-REVIEWED] 0ff5fce

Description read

[DESCRIPTION-READ] 80d326116ba4d6551e66bfd12a0368f4b07b29e8bfa033c6d41c099434d6d1d9

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 Oct 31, 2026 •

Copy link
Copy Markdown
Contributor

GPT 6.1 Review — ✅ no blocking findings

GPT 6.1 completed its review of 0ff5fcee27154e124a70221f6fd31decb75a2aa2 and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] 0ff5fce

False positive or not applicable? A repository writer can comment:
/ai-review override gpt 0ff5fcee27154e124a70221f6fd31decb75a2aa2 false-positive: <reason, 20 to 500 characters>

@github-actions

github-actions Bot commented Oct 31, 2026 •

Copy link
Copy Markdown
Contributor

Opus 5.5 Review — ✅ no blocking findings

Reviewed 0ff5fcee27154e124a70221f6fd31decb75a2aa2 — this comment is updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] 0ff5fce

Verdict parsed from the review's SHA-scoped output markers for commit 0ff5fcee27154e124a70221f6fd31decb75a2aa2.

False positive or not applicable? A repository writer can comment:
/ai-review override fable 0ff5fcee27154e124a70221f6fd31decb75a2aa2 false-positive: <reason, 20 to 500 characters>

@github-actions

github-actions Bot commented Oct 31, 2026 •

Copy link
Copy Markdown
Contributor

Security Scope Review -- ✅ PASS

Which legitimate operations does 0ff5fcee27154e124a70221f6fd31decb75a2aa2 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): 24 classified, 0 newly refused, 24 still allowed.
  • ubuntu-latest (platform posix): 24 classified, 0 newly refused, 24 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 only tightening in the shell composite is the new -c option-run reading in _nested_shell_payloads. I wrote 29 boundary candidates against it and none has a script-confirmed refusal. The new spec-shell-allowlist deny tier only acts on calls that carry a spec's stamped allowlist, and the classifier never stamps one, so I could not get a verdict for it. It is named under UNADJUDICATED. The curl -T/--upload-file rule deletion and the search/spec grant rows loosen the gate and are outside this lane.

TIGHTENING: nested-shell payload extraction (shell_normalizer._shell_option_run, which feeds the self-protection and git-publish floors) now walks the options and redirections after a shell's -c and adds the real script as a payload -- it refuses a protected command hidden behind options or redirections after -c, per + # A shell keeps reading options after ``-c``, so ``bash -c -x '<script>'`` runs ``<script>``` and + if j < _run_end(i):\n+ _c_script(j, k) # from the carrier: -co pipefail takes a name. Secondary, outside the shell composite: + GateTier("spec-shell-allowlist", "deny_policy", _tier_spec_shell_allowlist),`.

ADJUDICATED: 29 candidates, 0 confirmed newly-refused

PLATFORM COVERAGE: posix — 24 rows. They cover:

  • each option family the walk now consumes: -e, -x, -l, --, -o pipefail, -euo pipefail before -c, -oo in one cluster, -O extglob;
  • zsh/ksh's glued -onoclobber, read both the bash way and the getopt way;
  • the bash $0/$1 positional idiom, and positionals after the script;
  • redirections before and after the script, including a quoted target with a blank (the ambiguous-redirect emit);
  • carriers through $SHELL, /bin/bash, busybox sh, env, timeout, docker run and kubectl exec;
  • a grep -c inside the script;
  • the corpus's own push, commit, worktree, status and cron rows wrapped in bash -c <opts>.

windows — 5 rows: bash.exe by bare name, a quoted C:\Program Files\Git\bin\bash.exe path with a blank, MSYS2 sh.exe with \ separators, wsl bash, and a %TEMP%\ redirect target. These probe whether the walk's shell-program match (_program_basename against _NESTED_SHELL_PROGRAMS) and its redirect regex read the Windows spellings the same way. any — no rows, because every candidate names a shell executable. The diff also adds Windows path assumptions (_search_root_anchored's drive requirement, _glob_pattern_leaves_root's [\\/] split), but those sit in the read-only search grant: a loosening that no shell row can reach.

UNADJUDICATED:

  • Guard: _tier_spec_shell_allowlist (hook_runtime/gate_tiers.py), a deny_policy row placed after governance in GATE_TIERS.
  • Call path: the KAS session handle registers the agent spec. kas_permissions.spec_shell_allowlist(toolsSettings) sets AcpEvent.spec_shell_allowlist whenever toolsSettings.shell/execute_bash has denyByDefault. ToolCall.from_event → _allowlist_patterns turns that into the call's allowlist, then _command_matches_any decides. That function refuses any command that fails allowed_commands.plain_command, meaning one holding any of ;&|<>$\(){}\n\r or a control or non-space whitespace character, whatever the pattern says. It also refuses any pattern outside the literal-plus-.*` subset, and every pattern past the 64th.
  • Why there is no verdict: the deny composite the classifier runs never stamps a spec allowlist.
  • Narrowing argument I can make but not confirm: on kiro-cli, a denyByDefault spec with allowedCommands: ["npm run .*"] runs these commands, while Crew-on-KAS now refuses them on every approval policy, AUTO_APPROVE included:
    • npm run build 2>&1
    • npm run test | tee log
    • git status && git diff --stat
    • echo $HOME
    • a command with a tab
  • Candidate inputs I would have submitted:
    • for that spec: npm run build 2>&1, npm run lint -- --max-warnings 0
    • for a spec allowing pytest .*: pytest -q test\test_hooks.py (Windows backslashes)
    • for a spec allowing git .*: git log --format=%H -1
    • for a spec allowing [a-z]+ status, which is outside the subset: git status
    • a 65th listed command
  • The likely fix if any of these is confirmed: refuse only the chaining and substitution characters (;&| ` $( newline), not redirection or $VAR, or tell the operator at session start which listed commands will be refused.

[SCOPE-REVIEWED] 0ff5fce

@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 Oct 31, 2026
@bolichen97
bolichen97 force-pushed the backport/0.11.1-hotfix branch from 37649cc to 2049d09 Compare October 31, 2026 01:05
@github-actions github-actions Bot added readiness: checking Automated validation is still running readiness: passed Eligible automated validation passed for the current revision and removed readiness: action required A blocking check or review needs attention readiness: checking Automated validation is still running labels Oct 31, 2026

@KiroCrewCaptain KiroCrewCaptain left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

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

Approved on behalf of the watch holder (bolichen). Design and First Principles pass on this head, UX skipped or pass, no conflict, no open change request.

Backports these main PRs to release/0.11.0, in pick order. Each was a
`git cherry-pick -x` of its squash commit on main:

- fix(memory): read memory.db off the writer lock, split background pool (#24290) (cherry picked from commit fed830c)
- fix(dashboard): bound the pre-prompt build's wait to enter the memory pool (#24414) (cherry picked from commit 520cf2c)
- fix(memory): keep the shared -wal when another process closes the memory store (#25852) (cherry picked from commit dc63bb6)
- fix: read the page-cache prefetch in a child to keep sqlite locks (#26302) (cherry picked from commit 81dde23)
- perf(memory): score facts from a term index and prebuild after restart (#25328) (cherry picked from commit 6824348)
- fix(desktop): retry a failed netstat before Windows recovery gives up (#25792) (cherry picked from commit 043dd9f)
- fix(cli): say a gateway still serves the port when restart's replacement is refused (#25366) (cherry picked from commit 2e13127)
- fix(cli): refuse to stop or restart a gateway the desktop app supervises (#25705) (cherry picked from commit bbf0910)
- fix(hooks): approve kiro-cli read-only file searches after the deny tiers (#25150) (cherry picked from commit a0c0bd4)
- fix(settings): refuse a config save that config.local.json overrides (#26216) (cherry picked from commit 7f8688e)
- fix(workflows): write non-finite floats as null so runs open (#26085) (cherry picked from commit 3591282)
- fix(approvals): give each background tool approval its own id (#25905) (cherry picked from commit 51ef3bb)
- fix(chat): record file changes for kiro-cli v3 write tools (#25650) (cherry picked from commit 5369270)
- fix(sandbox): stop blaming a host-wide OOM kill on a per-tree limit (#25647) (cherry picked from commit 791d2a2)
- fix(acp): replay history when a KAS resume restores no turns (#25521) (cherry picked from commit 47c71cb)
- fix(kas): auto-approve granted built-ins after Crew's own checks (#25339) (cherry picked from commit dba4de5)
- fix(kas): mount web_search as the engine's remote_web_search (#26435) (cherry picked from commit 65c21b5)
- fix(kas): answer a spec's shell.allowedCommands at the gate, after Crew's checks (#25539) (cherry picked from commit 39d07d9)
- fix(kas): relay an MCP server's boolean env flags, never its secrets (#25331) (cherry picked from commit b734a68)
- fix(security): read the -c script past options and redirections (#26087) (cherry picked from commit fea0920)
- fix(kas): refuse a shell command a denyByDefault spec does not allow (#26501) (cherry picked from commit 9f0d02f)
- fix(kas): name a deniedCommands deny aimed at a redirect as not carried (#26489) (cherry picked from commit 4951c0c)
- fix(security): delete the curl-upload deny rule (#26469) (cherry picked from commit 70e206e)

Conflicts resolved by hand:
- #26302: test/spawn_audit_helpers.py adds only the page_cache.py entry
  (the embeddings_child.py entry names a module 0.11.0 lacks);
  test/test_page_cache_prefetch.py takes main's file.
- #25328: test/test_dashboard_server_composition_contract.py lazy-import
  count 93 -> 96 (the three names it adds).
- #25705: test/test_cli_restart_refused_message.py takes main's file.
- #26085: dashboard/handlers/workflows.py adds import math, keeps the
  release typing import.
- #25339: platform/governance.py keeps the release rule for a missing
  ceiling (main's enterprise-baseline clause is not on 0.11.0).
- #26435: harness-parity.md H6 row edited in place on the release text.
- #25539: its acp-client.md and agent-spec-fields.md lines are left out;
  the paragraphs they edit are not on 0.11.0 (security.md carries it).
- #25331: the test and two doc mentions use hoist_managed_servers /
  resume=, the 0.11.0 names of main's renamed hoist_servers_for_session.
- #26087: security package line budget 32,552 -> 32,714 (re-measured).
- #26501: kas_permissions.py comment keeps 0.11.0's shapes (main's
  subagent rules are absent); its acp-client.md line is left out, and the
  agent-spec-fields.md toolsSettings row says denyByDefault is gate-enforced.
- #26489: its agent-spec-fields.md sentence is added to the 0.11.0 row.
- #26469: test_denied_commands_security.py keeps 0.11.0's comment; count 113 -> 112.

Test fixes (host-dependent tests made hermetic; none loosens an assertion):
- tmpdir_helpers.short_tmp_path: a tmp_path override for modules whose code
  judges a path by its components (a .kiro/crew data-home part); an agent
  session's TMPDIR under ~/.kiro/crew made every test path the data home.
  Used by the file-explorer, design-tweak (both), spawn-offload, launch-copy,
  home-pin, native-skill and auto-improvement-members tests; the sage
  ancestor-chain test creates its dir under /tmp for the same reason.
- pi-gate tests and the remote-session Python 3.9 test take the first real
  binary on PATH (playwright_cli_installer_helpers._native_tool_on_path): a
  mise shim under a scratch HOME hangs or writes to stdout.
- snapshot-merge ACL test names its own uid; nobody (65534) is the overflow
  uid inside a user namespace and cannot be named in an ACL.
- installer-step and push_guard interrupt tests spawn with SIGINT at its
  default; a background pytest run starts with SIGINT ignored, which a shell
  trap and Python both keep.

Left out: #23920 (already on release/0.11.0), #26144 (needs
sandbox_launch_mark from #24709), #26088 (fences a main-only port),
#25346 (11 of its tests fail on 0.11.0: the kiro-cli skill scope it
compares against differs there), #26397 (needs #26152's new download
route, not on 0.11.0), #26428 (not merged).
@bolichen97
bolichen97 force-pushed the backport/0.11.1-hotfix branch from 2049d09 to 0ff5fce Compare October 31, 2026 03:38
@github-actions github-actions Bot added readiness: checking Automated validation is still running readiness: passed Eligible automated validation passed for the current revision and removed readiness: passed Eligible automated validation passed for the current revision readiness: checking Automated validation is still running labels Oct 31, 2026
@buluoray

Copy link
Copy Markdown
Contributor

Two candidates for this backport, both merged on main:

Happy to prepare the cherry-pick if that helps.

@bolichen97
bolichen97 merged commit ee5889b into release/0.11.0 Oct 31, 2026
34 of 38 checks passed
@bolichen97
bolichen97 deleted the backport/0.11.1-hotfix branch October 31, 2026 04:24
@github-actions github-actions Bot removed the readiness: passed Eligible automated validation passed for the current revision label Oct 31, 2026
bolichen97 added a commit that referenced this pull request Oct 31, 2026
…/0.11.0 (#26594)

Backports these main PRs to release/0.11.0 after #26487, in pick order.
Each was a `git cherry-pick -x` of its squash commit on main:

- fix(memory): keep the shared -wal on close when the driver has no setconfig (#26486) (cherry picked from commit ad856ad)
- fix(crew): blink a working icon's eyes and seat it flush in the crew stack (#25381) (cherry picked from commit d180166)
- fix(crew): counter-scale a working icon's frame in the face flight (#25020) (cherry picked from commit 94de6c6)
- fix(terminal): keep a newly opened terminal above the iPhone keyboard (#25370) (cherry picked from commit 670ab7d)
- fix(hooks): run agent spec hooks by kiro-cli's rules on KAS (#26428) (cherry picked from commit 314ea38)

Conflict resolved by hand:
- #26428: src/kiro_crew/hook_runtime/script_hooks.py keeps 0.11.0's
  extra_hooks_cwd branch (main passes **extra_run, which 0.11.0 lacks)
  and sets result.from_spec on both branches.

Left out (see the PR body): #19629, #23916, #23911 already on
release/0.11.0; #26001, #25419, #25142 need files only main has;
#26084 and #25483 use props and APIs only main has; #24634, #24635 and
#25364 conflict widely with code main changed after the cut; #26397
needs #26152's route; #26144 needs sandbox_launch_mark; #26456 not merged.

Co-authored-by: w <w@k.local>
iamwhatever pushed a commit that referenced this pull request Oct 31, 2026
….0 (#26627)

Backports these main fixes to release/0.11.0 after #26487 and #26594:

- fix(kas): let a relative file:// prompt use .. like kiro-cli (#26382) (cherry picked from commit 36c95db)
- part of #24709 (cherry picked from commit d86c94a, partial): the
  sandbox_launch_mark module, its note() call in cli.main before the
  KIROCREW_SANDBOX_ACTIVE pop, and its per-test reset fixture. Nothing
  else of #24709 (its enterprise policy-cache tier) is included; the
  module docstring and the cli.py comment name the 0.11.0 reader.
- fix(cloud): treat every agent process as an agent in the human-action guard (#26144) (cherry picked from commit dd0dd62)

Conflict resolved by hand:
- #26382: keeps 0.11.0's acp-client.md (the paragraph it edits is not on
  0.11.0); agent-spec-fields.md takes its path-fence bullets and keeps
  0.11.0's blank-prompt fallback bullet.

Test-only fixes:
- test_cloud_aws pins sandbox_launch_mark.SANDBOX_MARKER to
  sandbox_tiers._IN_SANDBOX_MARKER (the #24709 test that pinned it on main
  is not on 0.11.0).
- test_prompt_override_encoding pins browser_cli.install.available() off:
  the context build probed the host's PATH for playwright-cli and counted
  its refusal warning as a second override warning.

Left out: #26471 is a feature built on KAS code only main has.

Co-authored-by: w <w@k.local>
KiroCrewCaptain pushed a commit that referenced this pull request Oct 31, 2026
…NGELOG) (#26648)

Declare release/0.11.0 as the 0.11.1 hotfix line in the dual-valid RC
spelling the version numbering policy requires: all three version files
move from 0.11.0 to 0.11.1-rc.1. Add the ## [0.11.1] - 2026-10-31
section, written from the three hotfix backports already on the branch
(#26487, #26594, #26627). Every earlier section is unchanged.

Co-authored-by: w <w@k.local>
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