Release: version packages

[github-actions]

This PR was opened by the Changesets release GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated.

Releases

@tangle-network/browser-agent-driver@0.35.2

Patch Changes

• #129 c85666e Thanks @AtelyPham! - Fix the ffmpeg probe so a missing Playwright browsers cache disables video recording instead of keeping it on. The graceful degrade added in 0.35.1 only disabled recordVideo when the cache dir existed but had no ffmpeg; in the Tangle agent-thin sandbox PLAYWRIGHT_BROWSERS_PATH (/opt/cache/npm/_playwright) is never populated — Chromium is launched from Nix via executablePath — so the directory does not exist, and the previous "bias toward true on a missing dir" kept recordVideo and still crashed the run at context.newPage(). The probe now biases toward disabling video on any uncertainty (missing cache dir, cache without ffmpeg, or a probe error), because a wrong "available" hard-crashes the run while a wrong "unavailable" only skips the replay video; it returns true only when it positively finds an ffmpeg binary, or for the package-bundled PLAYWRIGHT_BROWSERS_PATH=0 mode. Normal dev/CI (where playwright install provides ffmpeg) is unchanged. The disabled-video warning now names the resolved cache path and the driver version.

[tangletools]

✅ Auto-approved GitHub Actions PR — db5e7a1f

This PR was opened by trusted GitHub Actions automation.
The full PR reviewer audit still runs separately and will publish findings if it detects issues.

_{tangletools · auto-approval · reason: github_actions_author · 2026-07-03T13:42:09Z}

[tangletools]

🟢 Value Audit — sound

Verdict	sound
Concerns	1 (1 weak-concern)
Heuristic	0.0s
Duplication	0.0s
Interrogation	51.2s (2 bridge agents)
Total	51.2s

💰 Value — sound

Version-bump release PR that publishes the 0.35.2 fix flipping the ffmpeg probe's failure bias from optimistic to pessimistic — a correct, well-reasoned asymmetry fix.

• What it does: This PR is the automated Changesets release vehicle: it bumps package.json from 0.35.1 to 0.35.2 (single line, package.json:4) so merging triggers OIDC npm publish. The substantive change it releases is in src/ffmpeg-availability.ts (commit c85666e): isPlaywrightFfmpegAvailable() now returns false (disable video) whenever ffmpeg cannot be positively confirmed — missing browsers cache dir, cache wi
• Goals it achieves: Stop the Tangle agent-thin sandbox from hard-crashing at context.newPage(). The sandbox launches Chromium from Nix via executablePath and never runs playwright install against PLAYWRIGHT_BROWSERS_PATH, so that cache dir does not exist; the 0.35.1 'bias toward true on a missing dir' kept recordVideo on and Playwright then threw 'Executable doesn't exist at .../ffmpeg-/ffmpeg-linux' killing t
• Assessment: Correct and in the codebase's grain. The 0.35.1 commit (fa9630c) introduced graceful degrade but had a real blind spot for the missing-dir case; this fix closes exactly that hole rather than reworking the mechanism. The asymmetry argument is sound and stated in the JSDoc (src/ffmpeg-availability.ts:38-58): a wrong 'available' hard-crashes the run, a wrong 'unavailable' only loses the replay video,
• Better / existing approach: none — this is the right approach. I verified the probe lives in the single dedicated module (src/ffmpeg-availability.ts) consumed only by src/cli/commands/run.ts:362, and grep for 'recordVideo' / 'ffmpeg' shows no parallel/duplicate detection logic elsewhere that this reinvents. It extends the existing 0.35.1 graceful-degrade seam rather than introducing a new one. The bundled-mode (=0) carve-out
• Model: opencode/zai-coding-plan/glm-5.2
• Bridge attempts: 2
• Bridge warning: opencode/kimi-for-coding/k2p7: bridge stream ended without value-audit content

🎯 Usefulness — sound

Release PR publishes the ffmpeg-cache fix (#129), which is correctly wired into both video-recording paths and soundly inverts the probe bias toward crash-avoidance.

• Integration: Fully reachable: isPlaywrightFfmpegAvailable is called from src/cli/commands/run.ts:360 (bad run) and src/showcase/index.ts:48 (bad showcase); the new playwrightBrowsersPath diagnostic helper is consumed at run.ts:362. Both are live on the real browser-launch path, not speculative.
• Fit with existing patterns: Extends the existing 0.35.1 graceful-degrade pattern (commit fa9630c) rather than competing with it — same probe, same call sites, just corrected bias. No parallel mechanism introduced.
• Real-world viability: Robust across error paths: missing cache dir -> false, cache-without-ffmpeg -> false, ENOTDIR (path is a file) caught -> false, probe exception -> false. Only positively-found ffmpeg or PLAYWRIGHT_BROWSERS_PATH=0 (package-bundled) yield true. The asymmetry rationale (wrong-available hard-crashes vs wrong-unavailable skips video) is correct and documented in-code.
• Model: opencode/zai-coding-plan/glm-5.2
• Bridge attempts: 1

🎯 Usefulness Audit

🟡 showcase warning not updated with cache-path/version detail [ergonomics] ``

run.ts:361-366 now names the resolved cache path and driver version in its disabled-video warning, but the parallel warning in src/showcase/index.ts:50-52 still uses the older generic message. Cosmetic inconsistency only — both paths disable video correctly; the showcase one is just less self-verifying. Optional: lift the richer message into the shared ffmpeg-availability module so both callers print the same diagnostic.

---

What this audit checks

It judges the change on its merits — not whether it was tasked out in an issue. Unticketed, fast-moving work is fine; the question is whether the change is good and whether a better or existing approach should be used instead.

Pass	What it asks
Heuristic	Vague title? Whitespace-only or cruft-bearing diff? (content signals only)
Duplication	Do added function/class names already exist elsewhere in the repo?
Value Audit	What does it do? What goal does it achieve? Is it good? Better architecture or already-exists?
Usefulness Audit	Does it integrate and fit? Will it hold up in real use and actually get used?

Findings are concerns, not blocks — the human reviewer decides what to do with them.

_{value-audit · 20260703T134509Z}

[tangletools]

✅ No Blockers — db5e7a1f

Review health 100/100 · Reviewer score 95/100 · Confidence 70/100 · 0 findings (none)

	glm	deepseek	aggregate
Readiness	95	95	95
Confidence	70	70	70
Correctness	95	95	95
Security	95	95	95
Testing	95	95	95
Architecture	95	95	95

Reviewer score is advisory once the run is complete and the verdict has no blockers.

Full multi-shot audit completed 2/2 planned shots over 2 changed files. Global verifier still owns final merge decision. | Full multi-shot audit completed 2/2 planned shots over 2 changed files. Global verifier still owns final merge decision.

No findings.

---

_{tangletools · 2026-07-03T13:45:41Z · trace}

[tangletools]

✅ Clean — db5e7a1f

Full multi-shot audit completed 2/2 planned shots over 2 changed files. Global verifier still owns final merge decision. | Full multi-shot audit completed 2/2 planned shots over 2 changed files. Global verifier still owns final merge decision.

Full immutable report for this review: trace

Summary comment for this run: full summary

---

_{tangletools · 2026-07-03T13:45:41Z · immutable trace}