| name | validate-and-fix |
| description | Run install sync, unit tests, and make verify; auto-fix trivial failures (ruff format/fix, obvious test expectation updates). Use when the user asks to validate, run tests, check the pipeline, or verify changes are clean. |
| disable-model-invocation | true |
Validate & Auto-Fix
Run the project validation pipeline, auto-fix trivial issues, and re-run until
green or a real failure is found.
Rules
- Never modify production logic to fix a test. Only fix test expectations,
imports, formatting, and lint.
- Never skip or delete a failing test.
- Stop after 3 auto-fix cycles to avoid loops.
- Report real failures clearly; do not attempt speculative fixes.
Step 0: Ensure Dependencies Are Installed
make install-all
Use make install if the user only needs dev tooling without all provider
extras. For most agent code paths, install-all matches CI-like coverage.
Step 1: Run Unit Tests
make test
If all pass, proceed to Step 3.
If failures occur, classify each failure (see Step 2).
Step 2: Classify and Fix Failures
Auto-fixable (fix immediately, then re-run Step 1):
| Type | Fix |
|---|
Missing await on async call | Add await, mark test async def, add @pytest.mark.asyncio |
| Test asserts old default value | Update assertion to match new default |
| Ruff format | make format |
Real failures (do not auto-fix without user):
- Logic errors in production code
- Assertion failures reflecting unclear behavior changes
- Import errors from missing optional extras (install with
make install-all)
Step 3: Run Verify
make verify
This runs: ruff format --check, ruff check, mypy on src/lightspeed_agentic.
If failures occur:
| Type | Fix |
|---|
| Format | make format |
| Unused import (F401) | Remove the import |
| Other ruff-fixable | make format (includes ruff check --fix) |
Re-run make verify after each fix.
Step 4: Report
Report exactly:
make test: pass/fail (counts if useful)make verify: pass/fail- Auto-fixes applied (file + what changed)
- Cycles used: N/3
Do not include unrelated diagnostics.
