molecule-ai/molecule-ai-plugin-molecule-skill-code-review
36
0
Fork 0
Code Issues Pull Requests Actions 22 Packages Projects Releases Wiki Activity
aeddebbfdd
molecule-ai-plugin-molecule.../tests
History
Molecule AI Plugin-Dev 8ec5295749
Some checks failed
CI / validate (push) Failing after 1s
Details
CI / validate (pull_request) Failing after 1s
Details
add smoke tests + rationale README 
All 5 thin-skill adapter repos now have:
- tests/README.md explaining limited coverage rationale (prose-only skills)
- tests/test_smoke.py covering plugin.yaml, SKILL.md frontmatter, adapter re-export

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-10 15:03:40 +00:00
..
README.md add smoke tests + rationale README 2026-05-10 15:03:40 +00:00
test_smoke.py add smoke tests + rationale README 2026-05-10 15:03:40 +00:00

README.md

Test Rationale — molecule-skill-code-review

What this plugin does

molecule-skill-code-review is a prose-only skill: its logic lives entirely in skills/code-review/SKILL.md. The adapter (adapters/claude_code.py) is a thin re-export of AgentskillsAdaptor from plugins_registry.builtins — no business logic, no network calls, no side effects.

What is tested

  • plugin.yaml is valid YAML with required fields (name, version, runtimes, skills)
  • skills/code-review/SKILL.md has valid YAML frontmatter and a body with headings
  • adapters/claude_code.py exists and re-exports AgentskillsAdaptor
  • validate-plugin.py (.molecule-ci/scripts/) exits zero

What is NOT unit-tested (and why)

The multi-criteria review logic (best-practices checks, modularity scoring, etc.) is prose guidance inside SKILL.md — there is no Python function to call. Testing it would require a full workspace runtime with a real Claude session. Smoke tests cover the artifact structure; full evaluation requires integration tests.

Running tests

python -m pytest tests/ -v