- Add a second monitored policy source only if it has a clean offline snapshot baseline and an equally honest governance story.
- Add one more supported procedure only if it can stay as deterministic and auditable as the current four.
- Tighten the rulebook workflow with a real draft snapshot only when there is a concrete candidate change worth reviewing.
- Add a lightweight screenshot capture path only if it stays reproducible and dependency-light.
- Structured upstream intake adapters.
- Human analyst annotation and review queues outside the deterministic core.
- More explicit governance dashboards or operational telemetry.
- Additional browser-level smoke coverage beyond Streamlit AppTest.
- LLM-first extraction
- approval prediction
- autonomous action
- a database-backed platform rewrite
- auth and deployment machinery
- broad procedure expansion without a defendable rule contract
Those ideas would widen scope without improving the validated deterministic core.