Improve RLCR efficiency with round typing, vertical-slice gates, and repeated-pattern checks

[VitalyAnkh]

Summary

A recent long RLCR session completed successfully and showed strong review rigor, but the analysis identified opportunities to reduce rework and improve progress per round.

The methodology worked well at preventing premature completion claims and catching concrete behavioral, verification, and integration gaps. The main improvement area is moving repeated-pattern detection, acceptance slicing, and loop-health checks earlier in the cycle.

Suggested Improvements

1. Add a mandatory round-type field:

   • mainline
   • review-fix
   • planning
   • recovery
   • tooling
   • documentation

2. Add a minimum vertical-slice requirement for mainline rounds:

   • behavior change
   • regression coverage
   • verification evidence
   • explicit exit condition

3. Add a repeated-pattern scan after each review finding:

   • identify adjacent surfaces where the same defect class may exist
   • fix adjacent occurrences or record why they are not applicable

4. Add a loop-health preflight:

   • test discovery works
   • review invocation works
   • minimal verification commands are available

5. Add pattern tags to review findings:

   • state-transition
   • partial-failure
   • boundary-consistency
   • verification-gap
   • user-feedback-mismatch

6. Add a no-silent-deferral rule:

   • acceptance-critical work cannot be moved to follow-up without explicit scope renegotiation

7. Add a surface-alignment gate when behavior changes:

   • user-facing instructions
   • configuration shape
   • typed models
   • validation commands

8. Add a round-end status line:

   • complete
   • incomplete
   • blocked
   • tooling-only

Expected Benefit

These changes should preserve RLCR review rigor while reducing repeated review-fix cycles and improving progress per round.