self-evolve

Audit and repair Claude configuration files, rules, and references.

113|15|Updated Mar 3, 2026
One-click install
npx skills add https://github.com/avibebuilder/claude-prime --skill self-evolve-avibebuilder
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: self-evolve
Source: https://github.com/avibebuilder/claude-prime/tree/main/.claude/skills/self-evolve
Command: npx skills add https://github.com/avibebuilder/claude-prime --skill self-evolve-avibebuilder

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This skill keeps a repository's Claude configuration coherent and up-to-date by detecting stale references, overlapping or weak rules, and noisy or duplicate self-improvement proposals so maintainers don't waste time on low-value edits or inconsistent agent behavior.

Core Features & Use Cases

  • Proposal consolidation: filter duplicates, merge related proposals, and batch-reject low-value items before presenting a focused set for approval.
  • Config audit and repair: analyze skills, rules, CLAUDE.md, hooks, and on-demand refs to find broken paths, trigger overlaps, and stale or misleading rules and apply safe fixes when appropriate.
  • Verification and reporting: smoke-test fixes, summarize what changed and where, and emit a health verdict for the repository's Claude setup.
  • Use case: run before a release or when onboarding contributors to ensure guards, skills, and references match the actual codebase.

Quick Start

Ask the skill to run a full audit and apply approved self-improvement proposals by saying "self-evolve" or "audit config".

Frequently Asked Questions about self-evolve

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I audit Claude config files for stale rules and broken references?

To audit Claude config files, run a repository-wide health check that analyzes .claude rules, CLAUDE.md, hooks, and on-demand references to detect broken paths, trigger overlaps, and stale or misleading rules, then apply safe fixes automatically.

What is the best way to consolidate self-improvement proposals in a Claude project?

Consolidating self-improvement proposals involves filtering duplicates, merging related items, and batch-rejecting low-value entries before presenting a focused set for approval, reducing noise and low-value edits for maintainers.

How do I check Claude agent trigger boundaries across skills and hooks?

Checking Claude agent trigger boundaries requires analyzing skills, rules, CLAUDE.md, and hooks to detect overlaps and verify trigger boundaries, ensuring guards and references match the actual codebase for consistent agent behavior.

Can I run a config health check before a release to verify Claude setup?

Yes, you can run a config health check before a release or when onboarding contributors to ensure guards, skills, and references match the actual codebase, generating a health verdict and summary of applied changes.

Do I need read access to .claude rules and hooks to audit project configuration?

Yes, auditing project configuration requires read access to .claude rules, CLAUDE.md, hooks, and the optional self-improve proposal store to detect broken references, overlaps, and weak rules effectively.

Why does my Claude agent behave inconsistently despite having rules defined?

Inconsistent Claude agent behavior often stems from overlapping rules, stale references, or broken paths in CLAUDE.md and hooks, which a config audit can detect and repair by applying safe fixes and smoke-testing the results.