[Spec 0041] E2E Test Suite for @cluesmith/codev

[waleedkadous]

Summary

• Add 70 BATS-based E2E tests that verify the npm package works after installation
• Tests cover all CLI commands: codev init, codev adopt, codev doctor, af, consult
• XDG sandboxing prevents pollution of dev environment during testing
• CI workflows for PR testing and post-release verification

Test Coverage

Test File	Tests	Coverage
install.bats	12	Package installation, binary availability
init.bats	14	Project initialization, structure creation
adopt.bats	11	Adding codev to existing projects
doctor.bats	8	Dependency checking
af.bats	11	Agent-farm CLI commands
consult.bats	14	Consultation CLI commands
Total	70

Test Execution

cd packages/codev
npm run build && npm pack
E2E_TARBALL=$(pwd)/cluesmith-codev-*.tgz bats ../../tests/e2e/

CI Workflows

• .github/workflows/e2e.yml - Runs on PRs (macOS + Linux matrix)
• .github/workflows/post-release-e2e.yml - Runs after npm publish

Test plan

• All 70 tests pass locally on macOS
• CI passes on both macOS and Linux
• Post-release workflow validated (manual trigger with version input)

[waleedkadous]

Architect Integration Review (3-Way Consultation)

Overall Verdict: APPROVE with minor feedback

Spec Update Required

The target duration in the spec should be updated from <2 minutes to <3 minutes. The ~2:45 actual time is acceptable given proper isolation (separate npm install per test).

Feedback to Address

1. Document the E2E_TARBALL pattern: Add a note in the test README or helpers explaining how to run tests locally with a custom tarball for debugging.

2. Version mismatch: The review noted CLI reports 1.0.0 but package.json has 1.1.0. While this is outside the e2e test scope, please add a TODO comment in the test noting this should be investigated separately.

3. Consider test parallelization for future: Not required now, but document in the review that parallel test execution could bring time under 2 minutes if needed later.

What Looked Good

• XDG sandboxing implementation is solid
• All 6 test categories (TC-001 through TC-006) covered
• Both CI workflows (PR and post-release) properly configured
• Error cases covered alongside happy paths

---

🏗️ Architect integration review

[waleedkadous]

Addressed all integration review feedback:

1. ✅ E2E_TARBALL pattern documented - Added usage examples in tests/e2e/helpers.bash header
2. ✅ Version mismatch TODO - Added comment in install.bats noting this should be investigated separately
3. ✅ Parallelization documented - Updated review with bats --jobs 4 recommendation
4. ✅ Target duration updated - Changed spec from <2 min to <3 min

Ready for final review.