Context
SPEC-02 defines which categories run on each cycle:
- Every cycle: Categories 1 (missing_tests) + 2 (stale_deps)
- Every 6 hours: Categories 3 (error_handling) + 4 (doc_drift) + 7 (security)
- Every 24 hours: Categories 5 (duplication) + 6 (performance)
- First ingestion only: Category 8 (eval_infra)
Current Behavior
AssessmentAgent.run() runs all registered modules on every call. No scheduling or rotation logic.
Expected Behavior
Track last run time per category. Only execute modules whose schedule interval has elapsed.
Impact
- Unnecessary work running expensive modules every cycle
- Not currently impactful since only 3 lightweight modules exist
- Will matter when LLM-dependent categories (3-7) are added
References
atlas-specs/02-ASSESSMENT.md — category rotation scheduleatlas/assessment/agent.py — runs all modules unconditionally
Context
SPEC-02 defines which categories run on each cycle:
Current Behavior
AssessmentAgent.run()runs all registered modules on every call. No scheduling or rotation logic.Expected Behavior
Track last run time per category. Only execute modules whose schedule interval has elapsed.
Impact
References
atlas-specs/02-ASSESSMENT.md— category rotation scheduleatlas/assessment/agent.py— runs all modules unconditionally