- tests/test_ecc_smoke.py: 23 tests covering: - plugin.yaml schema (5 skills, 3 rules, runtimes) - All 3 rules/ files exist and are non-empty - All 5 skills/ directories + SKILL.md with valid frontmatter + body heading - Claude Code + deepagents adapters exist and are wired - known-issues.md structure (severity definitions) - validate-plugin.py exit 0 smoke test - tests/README.md: explains why coverage is limited (skill+rules plugin, no executable hooks), and where to find integration-test guidance
37 lines
1.5 KiB
Markdown
37 lines
1.5 KiB
Markdown
# Test Coverage Rationale — molecule-ecc
|
|
|
|
## Why This Plugin Has Limited Unit-Test Coverage
|
|
|
|
`molecule-ecc` is a **skill+rules plugin** — it provides development guidelines and
|
|
development skills (api-design, coding-standards, deep-research, security-review, tdd-workflow)
|
|
via prose SKILL.md files and rules/*.md files.
|
|
|
|
There are no hooks, no Python business logic, and no testable adapters in this plugin.
|
|
The "logic" is prose documentation.
|
|
|
|
## What We Test (and Why)
|
|
|
|
| What | Why |
|
|
|------|-----|
|
|
| `plugin.yaml` schema | Verifies all 5 skills and 3 rules are registered |
|
|
| Rules files (3) | Each declared rule file exists and is non-empty |
|
|
| Skills (5) | Each skill directory + SKILL.md exists with valid YAML frontmatter and `#` heading |
|
|
| Adapters (2) | Claude Code + deepagents adapters are wired |
|
|
| `known-issues.md` | Severity definitions present |
|
|
| `validate-plugin.py` exit 0 | Smoke test — shared CI validator passes |
|
|
|
|
## What We Cannot Unit-Test Here
|
|
|
|
- **SKILL.md prose content** — the development guidelines are prose; their quality is
|
|
a documentation review concern, not a unit-test concern.
|
|
|
|
- **Agent behavior when using skills** — write integration tests in `workspace-template/`.
|
|
|
|
## Integration Tests
|
|
|
|
If you want to test that agents actually use the ecc skills correctly, write
|
|
integration tests that:
|
|
1. Install `molecule-ecc` on a test workspace
|
|
2. Ask the agent to use a specific skill (e.g., "use TDD workflow")
|
|
3. Verify the agent follows the documented process
|