src/__tests__/index.test.ts is described as an "integration test for the full run() pipeline", but it mocks @actions/core, node:fs, the parsers, the API client, and utils — so it verifies run() orchestration with everything stubbed, not real module integration.
True integration coverage lives in .github/workflows/e2e.yml (uses: ./ against fixtures) and the ci.yml self-test.
Suggested fix
- Update the description / CLAUDE.md wording so the test layers aren't misrepresented.
- Optionally add a thin real-integration test that exercises unmocked modules.
Minor.
Flagged during the AI-native docs revamp; tracked in .agents/known-issues.md (item F).
src/__tests__/index.test.tsis described as an "integration test for the fullrun()pipeline", but it mocks@actions/core,node:fs, the parsers, the API client, and utils — so it verifiesrun()orchestration with everything stubbed, not real module integration.True integration coverage lives in
.github/workflows/e2e.yml(uses: ./against fixtures) and theci.ymlself-test.Suggested fix
Minor.
Flagged during the AI-native docs revamp; tracked in
.agents/known-issues.md(item F).