# Evolving Solutions

> Recombine proven parts of multiple evaluated attempts into a coherent follow-up attempt. Use when ClimbHill attempt comparison reveals complementary partial wins.

- Skill: `closedloop-technologies/evolving-solutions` (Agent Skill)
- Install (CLI): `npx skillmds@latest add closedloop-technologies/evolving-solutions`
- Raw SKILL.md: https://api.skillmd.com/api/skills/closedloop-technologies/evolving-solutions/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: closedloop-technologies (https://skillmd.com/u/closedloop-technologies)
- Updated: 2026-09-10
- Page: https://skillmd.com/skills/closedloop-technologies/evolving-solutions

---


# Evolving Solutions

Use this skill when combining lessons from multiple attempts into a stronger follow-up attempt.

## Inputs

- Current goal
- Attempt comparison report
- Attempt diffs or summaries
- Evaluation results
- Policy outcomes

## Procedure

1. Identify the strongest attempt and the specific reasons it is strongest.
2. Identify partial wins from weaker attempts.
3. Exclude changes that failed policy, broke tests, or contradicted human decisions.
4. Create a recombination plan that lists what to keep, discard, and rewrite.
5. Implement the smallest coherent attempt that combines the useful ideas.
6. Run required checks and record lineage as `combined_with` or `inspired_by`.

## Outputs

- Recombination plan
- New attempt summary
- Lineage relationships
- Evaluation results

## Rubric

- The combined attempt is simpler than the union of all attempted patches.
- Known failures are not reintroduced.
- Lineage is explicit.
- The attempt can be reviewed as a normal patch.