Self Improvement
This is meta documentation for Claude itself on how to self-improve by studying real projects like actionbook and accomplish-ai. The core insight is token efficiency: don't read entire files when you need one section, pre-compute patterns instead of guessing at runtime, and maintain reusable skills like action manuals. It includes concrete patterns like layered architecture, security defaults, and a checklist for when to auto-install tools or create new skills. Honestly, this reads more like internal agent guidelines than something most users would directly invoke, but if you're building autonomous AI systems or want Claude to proactively maintain its own tooling, the frameworks here are solid.
- skills.sh installs
- 167
- last updated
- Aug 21, 2026
Install
$ npx -y skills add undermybelt/self-improvement-skill --skill self-improvement --agent claude-codeInstalls into .claude/skills of the current project.
Files
Self-Improvement Patterns for AI Agents
Patterns learned from studying actionbook and accomplish-ai repositories.
Core Principles
1. Token Efficiency is Critical
| Approach | Tokens | Cost |
|---|---|---|
| Full HTML page | 50,000 | $$$ |
| Relevant DOM only | 500 | $ |
| Savings: 100x |
Apply to everything:
- Don't read entire files when you only need a section
- Use semantic search (QMD) instead of grep through everything
- Batch related operations
2. Pre-compute > Runtime Guess
Actionbook's key insight:
- Pre-compute DOM selectors → Store as "action manuals"
- Agent looks up manual → Knows exactly what to do
- No guessing, no hallucination
Apply to your work:
- Cache frequently used patterns
- Store configurations, not just code
- Build "manuals" for complex operations
3. Layered Architecture
From Accomplish:
packages/shared/ ← Types, constants (no dependencies)
packages/core/ ← Business logic (depends on shared)
apps/desktop/ ← UI/Electron (depends on core)
Sponsors
Agent CaffeineSponsoredClose your lid, come back to finished work
SnugSponsoredThe clipboard manager for the AI eraPromote your product hereReach 125k+ visitors monthly
WaxSponsoredBlock anything with supervision, without erasing your iPhonePromote your product hereReach 125k+ visitors monthly
knobsSponsoredSee your site on any phone, any day, in any languageDetails
- Repository
- github.com/undermybelt/self-improvement-skill
- First seen
- Jun 3, 2026
Recommended
cavemanjuliusbrussee/cavemanUltra-compressed communication mode. Cuts output tokens 65% (measured) by speaking like caveman while keeping full technical accuracy. Supports intensity levels618.3Kskills.sh installs
improveshadcn/improveSurvey any codebase as a senior advisor and produce prioritized, self-contained implementation plans for OTHER models/agents to execute. Strictly read-only on s40Kskills.sh installs
systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixes326.7Kskills.sh installs
karpathy-guidelinesforrestchang/andrej-karpathy-skillsBehavioral guidelines to reduce common LLM coding mistakes through explicit assumptions, simplicity, and verifiable success criteria.17.9Kskills.sh installs
find-skillsvercel-labs/skillsHelps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express int4.3Mskills.sh installs