subhashmahimaluri/Vite PR Coverage Insight

Compare coverage between main and PR, post result to PR comment

View on GitHub

Trust Signals

Scorecard Score
not yet scored
Maintenance Recency
Stale
License
None
namedescriptionrequireddefault
github-tokenGitHub token for PR commentyes
mode'report' (default) posts the PR comment; 'baseline' publishes coverage to the baseline store on main-branch pushesnoreport
basePath to base coverage-summary.json (optional since v2 — omit to auto-resolve from the baseline store)no
headPath to PR coverage-summary.json, or a directory of shard summaries to mergeno
coveragePath to coverage-summary.json for 'baseline' mode (falls back to 'head')no
baseline-branchOrphan branch used as the baseline/history storenocoverage-baseline
baseline-modeHow to resolve the base coverage: 'auto' (recorded baseline, falling back to dual-run), 'branch' (recorded only), 'scan' (dual-run: test the merge-base in this job — no baseline branch needed, requires checkout fetch-depth: 0 and run-script), or 'off'noauto
base-run-scriptCommand run on the merge-base worktree for dual-run baselines (defaults to run-script). Include dependency install, e.g. 'npm ci && npm run test:coverage'no
fix-plan-fileMarkdown coverage fix plan — changed files with gaps first, one ready-to-paste AI test prompt each (Copilot Chat etc.)nocoverage-fix-plan.md
mutationPath to a Stryker mutation report (json reporter output, mutation-testing-report-schema). Auto-detected at reports/mutation/mutation.json when omitted. Adds the 🧬 Mutation score card to the band, lists surviving mutants in changed files in the comment, and adds kill-prompts to the fix plan. In 'baseline' mode the score is recorded so PR runs show a Δ.no
upload-artifactUpload JSON + HTML + fix plan as a run artifact and link it from the PR comment (📥 download the full report)notrue
cacheActions cache acceleration for baseline lookups: 'on' (default) or 'off'. The baseline branch remains the source of truth, so 'off' only costs one extra API read — use it if the cache service is flaky on your runners.noon
thresholdsCoverage gate enforced by the action: a number for all metrics ('80'), per-metric pairs ('lines:85, branches:75') or JSON. Overrides coverage-insight.config.json. When unmet the job fails with a compliance table; without it the action only reports.no
aiAI assistance: 'off' (default, zero non-GitHub network calls), 'comment' (risk analysis + test suggestions), 'review' (adds advisory verdict)nooff
ai-can-blockOnly with ai: 'review' — allow a high-risk AI verdict to set the check run to neutralnofalse
run-scriptShell command to run tests with coverage before reading the coverage file (e.g. npm run test:coverage). On a non-zero exit the report still posts: failed test names are parsed from the output (vitest/jest), the comment enters the tests-failed state and the job is marked failed so the PR cannot merge. A non-zero exit caused only by the runner's own coverage thresholds is a warning, not a failure — use the `thresholds` input to gate.no
test-failuresPath to test failures JSON fileno
use-check-runWhether to use GitHub Check Run APInofalse
annotationsDiff annotations published via the check run: 'all' (default), 'coverage' (uncovered lines in touched files), 'failed-tests', or 'none'noall

no outputs