| name | self-assess |
| description | Analyse the codebase and BDD coverage to find gaps, bugs, and improvement opportunities |
| tools | ["bash","read_file"] |
| pipelines | ["evolve","claude"] |
Self-Assessment
Process
- Read BDD.md — know the full spec before assessing anything
- Read BDD_STATUS.md — see which scenarios are covered and which aren't
- Read the source code — does the implementation actually match the scenarios?
- Try running the tests — are they all passing? Any flakes?
- Check JOURNAL_INDEX.md — quick scan of past sessions. Read JOURNAL.md only if you need detail on a specific day.
What to look for
- Scenarios in BDD.md with no corresponding test (UNCOVERED)
- Tests that exist but are wrong — they pass but don't actually test the scenario
- Code that implements behaviour not described in any scenario (dead code, or missing BDD coverage)
- Build failures, linting errors, type errors
- Edge cases within a scenario that aren't handled
BDD coverage check
Run this to see the current state:
python3 scripts/check_bdd_coverage.py BDD.md
Any scenario marked UNCOVERED is your highest-priority work item.
Output format
Write findings as a prioritised list:
ASSESSMENT Day [N]:
1. [CRITICAL] Scenario "X" is UNCOVERED — no test exists
2. [HIGH] Test for "Y" passes but doesn't actually verify the Then clause
3. [MEDIUM] Code in auth.py has no corresponding BDD scenario
4. [LOW] Lint warning in utils.js
Pick the highest-priority item and implement it this session.