[Fooftilly]

Summary

Second B1 informational coverage slice under #651, following the merged Python coverage pilot #662.

Adds one path-scoped GitHub Actions workflow to measure line and branch coverage for frontend-app/src/lifecycle/tab-leave.ts using its existing tab-leave.test.ts Vitest suite. This module implements owner freshness, cancellation/rejection, flush and commit ordering, and batch atomicity; it is more safety-critical than UI rendering/style code.

Design

• Uses the existing Vitest 5.0.2 from frontend-app/package-lock.json and @vitest/coverage-v8@5.0.2, matching the pinned version exactly.
• Adds @vitest/coverage-v8@5.0.2 as a pinned development-only dependency in frontend-app/package.json, its generated npm lockfile, and the accepted dependency-inventory.json (required by PRKS's dependency preflight), matching vitest@5.0.2. CI installs everything with only npm ci --ignore-scripts; no ad-hoc installation, unpinned package, production dependency, or generated bundle is introduced. This addresses the zizmor unlocked-install finding and Codex's unlocked-transitive-dependency finding on the initial pilot.
• Runs only the existing src/lifecycle/tab-leave.test.ts suite with V8 coverage, restricting the report to src/lifecycle/tab-leave.ts. It does not rerun all frontend tests or modify the main Vue CI job.
• Writes separate line and branch covered/total values and percentages to the GitHub Actions job summary; json-summary stays in RUNNER_TEMP, never uploaded. Validates both denominators are nonzero to prevent falsely reporting an empty pilot.
• Informational only: no coverage thresholds, fail-under, PR comments, or coverage artifact uploads. Failing tests or a broken reporter still fail the pilot job.
• PR-triggered only for the workflow, targeted source/test, Vue test config, or npm manifests; manual workflow_dispatch is also supported. Minimal read-only permissions, pinned checkout/setup-node action SHAs, Node version consistent with the main Vue check.

Validation / acceptance

• Base: master@104e1499f74b5522f8795fad906a731367a20f0d (merge of ci: pilot informational Python line and branch coverage (#651) #662).
• Verify the pilot runs and produces actual nonempty per-module line/branch totals; inspect real CI results, not assumed percentages.
• Verify all regular CI checks, actionlint and zizmor, as well as review findings. If provider installation exposes a problem, fix the pilot rather than weakening existing gates.
• Verified lockfile-backed coverage baseline: run 38557111143 passed 12 tests, reporting 77/94 lines (81.9%) and 53/84 branches (63.1%) in tab-leave.ts only. The final head 782baca has now passed all eight GitHub Actions workflows, including the inventory-enforcing PRKS Test Gate, static analysis, CodeQL, actionlint, zizmor, Dependency Review, the Vitest pilot, and the Full E2E Gate. The original E2E attempt failed once on a Work Reminders Playwright wait; failed-job rerun (attempt 2) passed all six E2E shards, including the previously failing test. No unrelated product changes were made.
• Record observed CI cost and coverage baseline in this PR. Further B1 work (broader modules, meaningful changed-line metrics, trend tracking) remains separate and nonblocking.

Scope: One new workflow plus one pinned dev dependency, its npm lockfile and the required dependency inventory entry; no runtime, PostgreSQL, offline-state, or existing CI gate changes. No automatic merge; leave open for maintainer approval.

Related: #651 (B1), #662, and tab leave ownership #233.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	🔄 Running since 2026-10-14T11:33:50.608895Z	782baca	New commits

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[Fooftilly]

One finding on the provider install step (inline). The pilot itself ran and produced real totals (12 tests passed; Lines 77/94, 81.9%; Branches 53/84, 63.1%).