← ClaudeAtlas

run-checkslisted

Local "run everything CI would run" entrypoint — validates state.json against state.schema.json, runs the tooling's own tests/ suite, and runs pytest (or the configured test_command) across every generated day's and micro-project's code/tests/. Use when the user invokes /run-checks, or before anything that should be shipped/committed as a batch.
hariomlohardev/claude-system · ★ 1 · Data & Documents · score 53
Install: claude install-skill hariomlohardev/claude-system
# /run-checks Read-only with respect to `state.json` (it only validates, never edits it). This is the same thing `.github/workflows/ci.yml` runs automatically on push — this command lets the user run it locally on demand, and it's also what `/safe-revert` should suggest running after a revert to confirm the repo landed in a good state. ## Steps 1. **Schema check**: run `python3 scripts/validate_state.py`. Report pass/fail clearly. If `state.json` doesn't exist yet, note that plainly (not an error — just means no month has been started). 2. **Tooling test suite**: run `pytest tests/ -v` (this exercises `scripts/validate_state.py` itself against the fixtures in `tests/fixtures/`). Report pass/fail per test, not just a total. 3. **Generated content tests**: for every `month_<N>/week_<W>/day_<D>/code/tests/` and `.../micro_projects/<slug>/code/tests/` that exists on disk, run the appropriate test command: - `months.<N>.language == "python"` (or unset/default): `pytest <path> -v`. - `"javascript"`/`"typescript"`: `npm test` from that day's `code/` directory, if a `package.json` is present there. - `"other"`: `months.<N>.test_command`. Report each day/micro-project's result separately — don't just give one aggregate pass/fail, since a single failing day shouldn't obscure that everything else is fine. 4. **Summary**: a short pass/fail table — schema check, tooling suite, and one line per day/micro-project tested. If everything passed