self-improving-agent

Analyze Claude Code auto-memory patterns and promote proven learnings into reusable skills.

Updated Apr 2, 2026

Search Tags:#Claude Code #memory management #pattern analysis #project knowledge #skill promotion

Ready to run

npx skills add https://github.com/4lerman/text_evaluator --skill self-improving-agent-4lerman

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: self-improving-agent
Source: https://github.com/4lerman/text_evaluator/tree/main/.agents/skills/engineering-skills/self-improving-agent
Command: npx skills add https://github.com/4lerman/text_evaluator --skill self-improving-agent-4lerman

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) and agents (resource) components.

What problem does it solve?

The Self-Improving Agent addresses the issue of managing Claude Code's auto-memory by curating it into durable project knowledge, analyzing patterns, and promoting proven learnings.

Core Features & Use Cases

Memory Analysis: Analyze auto-memory for promotion candidates, stale entries, consolidation opportunities, and health metrics.
Promotion: Gradually promote patterns from auto-memory to CLAUDE.md or .claude/rules/ for permanent enforcement.
Skill Extraction: Transform proven patterns into reusable skills with SKILL.md, reference docs, and examples.
Use Case: When you have a large number of project learnings and want to maintain them effectively without cluttering your project files.

Quick Start

To analyze the auto-memory for patterns, run the command /si:review. To promote a learning to a rule, use /si:promote <pattern description>.

Frequently Asked Questions about self-improving-agent

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

FAQPage Schema

How do I manage Claude Code auto-memory to prevent clutter in my project files?▼

You can manage Claude Code auto-memory by curating learnings into durable project knowledge, analyzing patterns, and promoting proven entries to rules. This prevents clutter by transforming scattered memories into structured, reusable skills.

How does promoting a pattern to CLAUDE.md work for project knowledge?▼

Promoting a pattern to CLAUDE.md involves analyzing auto-memory for proven learnings and gradually enforcing them as permanent rules. This transforms temporary project knowledge into durable, reusable skills for consistent application.

What is the best way to extract reusable skills from auto-memory?▼

The best way to extract reusable skills from auto-memory is to analyze patterns for promotion candidates and transform proven learnings into structured formats with reference docs and examples. This ensures extracted skills are durable and effective.

Can I analyze auto-memory for stale entries and consolidation opportunities?▼

Yes, you can analyze auto-memory for stale entries, consolidation opportunities, and health metrics. This review process identifies which project learnings are outdated and which patterns are ready for promotion to permanent rules.

When do I need to curate auto-memory into durable project knowledge?▼

You need to curate auto-memory into durable project knowledge when you have a large number of project learnings and want to maintain them effectively. This prevents auto-memory from becoming unmanageable and ensures proven patterns are permanently enforced.