iterative-improvement - Iteratively improve open-ended work with measurable review
Runs a criteria-driven loop of hostile review, triage, execution, and independent scoring to improve work without a single objective completion test.
Tags
Updated: 2026-10-08Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Interview users
- Define weighted criteria
- Save criteria documents
- Run hostile reviews
- Triage review findings
- Execute prioritized changes
- Score progress independently
- Report loop results
Inputs
- User goals and preferences
- Existing artifact
- Repository context
- Review constraints
- Stop condition and round budget
- Off-limits scope
Outputs
- Criteria document
- Review findings
- Prioritized changes
- Score log
- Improved artifact
- Final report
Requirements
- Writable repository access
- Access to criteria guide
- User approval of criteria
- Permission to modify artifacts
- Independent reviewer
- Independent scorer