Skip to content

fix(browser): 0.13 pick of #39267, reinstall the engine by itself after an update - #39476

Merged
buluoray merged 1 commit into
release/0.13.0from
crew/slack/pick-39267-0.13
Nov 19, 2026
Merged

buluoray merged 1 commit into
release/0.13.0from
crew/slack/pick-39267-0.13

Conversation

@buluoray

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: 0.13 users get the browser engine reinstalled by itself after an update, without clicking Install.

This is the 0.13 pick of #39267, the user-reported sibling of #39265 (which reached 0.13 through #39387). After every update, Settings > Browser showed Install with no reason, and the agent's browser tool said browsing is not set up.

Why it matters

Anyone on a packaged release that keeps each version in its own directory loses the browser after an update until they click Install. 0.13 is the release where that experience should be fixed, and #39265's fallback fix is already on it.

Not a goal

What changed (motivation → approach → change)

git cherry-pick -x c8ff0adb of #39267 onto release/0.13.0. Every file applied cleanly except one hunk in src/kiro_crew/dashboard/handlers/browser_ops.py. On main that hunk sits in run_browser_op, which #39022 added; release/0.13.0 still has the route-level api_browser_op. The hunk is hand-ported to that shape: it reads the dashboard state from request.app.get("state") and answers with web.json_response, which is the form #39267 had before main's refactor. The repair logic, the browser_repairing answer and the tests are the same as on main.

Part On 0.13
repair_browser_harness and the startup re-check (browser_harness_status.py) same as main
op answer browser_repairing (browser_ops.py) hand-ported to the pre-#39022 route
tests and browser module spec same as main

Backwards compatibility

Compatible: only a home whose install broke after startup behaves differently, and that home could not browse before.

Removes nothing: no key, route or field is removed or narrowed.

Tests

release/* gets no backend test lanes in CI, so the related files ran locally on this branch, single process (-n0): the five files #39267 touches or guards (123 passed), then every test/test_browser*.py plus test/test_mcp_browser_tool.py in two sequential batches (865 and 841 passed, 1 skipped).

Proof: not provable -- other: a pick of #39267, whose Proof line on main was PROVEN at head c435961

Manual verification

N/A — same change as #39267 on main; unit coverage drives the dangling-link state.

Real-machine verification

N/A -- a pick; the base behaviour was checked on a packaged release for #39267 and a packaged 0.13 build of this branch is not available from this setup.

Related Issues

Part of #39254

Sibling sweep

Siblings: none -- this is a pick of one main fix; #39267's own sweep covered the browser handlers.

Pattern harvest

Not generalizable: a release pick of one fix whose pattern harvest is on #39267.

Checklist

  • A Conventional Commits title (feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...); PR Hygiene does not count commits
  • 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

…ts interpreter (#39267)

(cherry picked from commit c8ff0ad)

# Conflicts:
#	src/kiro_crew/dashboard/handlers/browser_ops.py
@buluoray
buluoray requested review from a team and KiroCrewCaptain as code owners November 19, 2026 13:31
@buluoray
buluoray requested review from bolichen97 and removed request for a team November 19, 2026 13:31
@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 58ad666f8560669d53a720cc318289de9759c4f7 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

Design Review (Opus 5.5) — ✅ PASS

Design-level review of 58ad666f8560669d53a720cc318289de9759c4f7 — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS does not.

Design-Verdict: PASS

Nothing blocks: this is a faithful 0.13 pick of #39267, the one hand-ported hunk keeps the same repair semantics, and nothing is taken away.

[DESIGN-REVIEWED] 58ad666

Description read

[DESCRIPTION-READ] 1744fba2f7a01291a73b1fc99562ad36ffadc6ffa35be0d7c1681b2aa1a666fd

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

First Principles Review (Opus 5.5) — ✅ PASS

Premise-level review of 58ad666f8560669d53a720cc318289de9759c4f7 — 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 release pick of #39267. It finishes the startup auto-install that already ships by covering an interpreter removed later, so no new product shape needs a decision.

Notes

  • Item 4 — one consumer, generalized: repair_browser_harness(..., supervisor=None, environ=None). Production code never passes environ (2 callers, both omit it), and the browser_ops.py caller omits supervisor as well. Only tests use them. Subtraction: none, because this is a pick and its Not-a-goal is "Any change beyond fix(browser): reinstall the engine by itself when an update removes its interpreter #39267's own diff". Shrink it on main instead.

What this change ships

Inventory (6 items) — 5 justified

Intent: after an update deletes the venv's interpreter, 0.13 users get the browser engine reinstalled without clicking Install. This is a FIX.

  1. Opening Settings > Browser on an install whose interpreter is gone starts the reinstall by itself — justified
  2. An agent's browser op on that install starts the reinstall and gets the answer "being installed; try again in a minute" — justified
  3. Any browser op while an install job runs (including an Install press) answers browser_repairing instead of not_installed — justified
  4. A new public repair_browser_harness, limited to once per process, never over a running or failed job, and skipped under test/CI — one consumer, generalized: environ is only a test seam
  5. The startup auto-install now leaves alone a job that a repair or an Install press started during discovery — justified
  6. A spec paragraph in browser.md describes the repair — justified

[FIRST-PRINCIPLES-REVIEWED] 58ad666

Description read

[DESCRIPTION-READ] 0d7b8f7fd0fab658716935e3bc0c7fa9354ed37af649d22eceb6af3ce54dcba8

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

Opus 5.5 Review — ✅ no blocking findings

Reviewed 58ad666f8560669d53a720cc318289de9759c4f7 — this comment is updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] 58ad666

Verdict parsed from the review's SHA-scoped output markers for commit 58ad666f8560669d53a720cc318289de9759c4f7.

False positive or not applicable? A repository writer can comment:
/ai-review override fable 58ad666f8560669d53a720cc318289de9759c4f7 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 58ad666f8560669d53a720cc318289de9759c4f7 and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] 58ad666

False positive or not applicable? A repository writer can comment:
/ai-review override gpt 58ad666f8560669d53a720cc318289de9759c4f7 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 19, 2026
@buluoray
buluoray merged commit 7a151b9 into release/0.13.0 Nov 19, 2026
90 checks passed
@buluoray
buluoray deleted the crew/slack/pick-39267-0.13 branch November 19, 2026 14:54
@github-actions github-actions Bot removed the readiness: passed Eligible automated validation passed for the current revision 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