self-improvement
Analyze conversation history to find friction patterns and suggest CLAUDE.md/skill improvements. Use when user wants to review what went wrong across sessions and systematically improve. (user)
Best use case
self-improvement is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
Analyze conversation history to find friction patterns and suggest CLAUDE.md/skill improvements. Use when user wants to review what went wrong across sessions and systematically improve. (user)
Teams using self-improvement should expect a more consistent output, faster repeated execution, less prompt rewriting.
When to use this skill
- You want a reusable workflow that can be run more than once with consistent structure.
When not to use this skill
- You only need a quick one-off answer and do not need a reusable workflow.
- You cannot install or maintain the underlying files, dependencies, or repository context.
How self-improvement Compares
| Feature / Agent | self-improvement | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | Unknown | N/A |
Frequently Asked Questions
What does this skill do?
Analyze conversation history to find friction patterns and suggest CLAUDE.md/skill improvements. Use when user wants to review what went wrong across sessions and systematically improve. (user)
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
SKILL.md Source
# Self-Improvement - Learn from History Analyze Claude Code conversation history to find friction patterns, check what's already fixed, and suggest improvements. ## Instructions ### Phase 1: Find Conversations ```bash # Get project directory (encode current path with dashes) ls ~/.claude/projects/ # List today's conversations (or adjust -mtime for longer range) ls -lt ~/.claude/projects/<encoded-path>/*.jsonl | head -20 ``` ### Phase 2: Parallel Analysis Spawn Task agents (subagent_type: general-purpose) to analyze conversations in parallel. Each agent reads one .jsonl file and extracts: 1. What was user trying to accomplish? 2. Problems/friction that occurred (include user quotes showing frustration) 3. What worked well? 4. Repeated patterns or inefficiencies For large files (>500KB), prioritize those - they contain the meatiest sessions. ### Phase 3: Synthesize After all agents complete, combine findings: - **Top friction patterns** ranked by frequency - **What worked well** (don't lose these) - **User frustration quotes** (raw evidence) **Generalize aggressively.** Look for the meta-pattern behind specific issues. ### Phase 4: Cross-Reference Read current documentation: - Project CLAUDE.md - User ~/.claude/CLAUDE.md - Skills in .claude/skills/ and ~/.claude/skills/ For each friction pattern, classify: - **Already fixed** - note location - **Still missing** - needs addition ### Phase 5: Output Create a markdown file (e.g., `CLAUDE_IMPROVEMENTS.md`) with: ```markdown # Suggested CLAUDE.md Improvements Based on analysis of N conversations from [date range]. ## 1. [Issue Name] **Problem:** [Description of friction] **Suggested addition to [location]:** \```markdown [Proposed text] \``` --- ## Already Fixed | Issue | Where | |-------|-------| | ... | ... | --- ## Potential Skills | Skill | Purpose | |-------|---------| | ... | ... | --- ## Raw Friction Log - "user quote 1" - "user quote 2" ``` **Do not apply changes** - create the file for user review. ## Usage Examples ``` /self-improvement /self-improvement last 3 days /self-improvement refactoring ```
Related Skills
self-improvement
Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Claude ('No, that's wrong...', 'Actually...'), (3) User requests a capability that doesn't exist, (4) An external API or tool fails, (5) Claude realizes its knowledge is outdated or incorrect, (6) A better approach is discovered for a recurring task. Also review learnings before major tasks.
self-xyz
Integrate Self (self.xyz) — a privacy-first identity protocol using zero-knowledge proofs to verify passports and ID cards. Use when the user mentions Self protocol, Self identity, self.xyz, passport verification, zero-knowledge identity verification, SelfAppBuilder, SelfBackendVerifier, SelfVerificationRoot, or wants to add privacy-preserving KYC, age verification, nationality checks, OFAC screening, or Sybil resistance using real-world identity documents. Covers frontend QR code integration, backend proof verification, and on-chain smart contract verification on Celo.
mobile-self-test
Self-test a mobile feature change or bug fix after implementation in `apps/mobile`. Use this whenever the user asks to verify a mobile change, run simulator acceptance, smoke-test a mobile PR, or provide screenshot proof for a mobile fix. This skill decides between prod vs local API mode, starts the local follow-server when needed, builds a release app, uses Maestro only to bootstrap registration for non-auth work, then switches to screenshot-driven visual validation and returns screenshot evidence.
self-improving-agent
Curate Claude Code's auto-memory into durable project knowledge. Analyze MEMORY.md for patterns, promote proven learnings to CLAUDE.md and .claude/rules/, extract recurring solutions into reusable skills. Use when: (1) reviewing what Claude has learned about your project, (2) graduating a pattern from notes to enforced rules, (3) turning a debugging solution into a skill, (4) checking memory health and capacity.
self-eval
Honestly evaluate AI work quality using a two-axis scoring system. Use after completing a task, code review, or work session to get an unbiased assessment. Detects score inflation, forces devil's advocate reasoning, and persists scores across sessions.
auto-paper-improvement-loop
Autonomously improve a generated paper via GPT-5.4 xhigh review → implement fixes → recompile, for 2 rounds. Use when user says "改论文", "improve paper", "论文润色循环", "auto improve", or wants to iteratively polish a generated paper.
implementing-runtime-application-self-protection
Deploy Runtime Application Self-Protection (RASP) agents to detect and block attacks from within application runtime, covering OpenRASP integration, attack pattern detection, and security policy configuration for Java and Python web applications.
metacognitive-self-mod
Analyze and improve the improvement process. Use for detecting quality regressions and refining meta-optimization
self-optimization
Turn mistakes, corrections, dead ends, and repeated fixes into durable improvements. Use when work reveals a non-obvious lesson, a recurring failure, a missing capability, or a rule that should be promoted into agent memory, workflow guidance, or a reusable skill.
openclaw-self-guard
OpenClaw Security Vulnerability Monitor - Checks for OpenClaw security vulnerabilities from NVD CVE database and GitHub Security Advisories. Compares local OpenClaw version against known CVEs, outputs vulnerability details and remediation if found, otherwise runs silently. Auto-installs daily cron job. Data sources: NVD (nist.gov), GitHub Security Advisories.
criticism-self-criticism
触发：当一项工作已经完成、进入阶段验收、收到批评反馈，或反复出现同类错误需要系统纠偏时调用；常见信号包括 review、audit、retrospective、quality check、纠错与复盘。 English: Trigger after delivery or at a review checkpoint when quality must be examined honestly and errors must be corrected without defensiveness. Use this skill for structured self-review, feedback processing, and continuous correction.
agent-self-repair
General AI agent introspection debugging framework: auto capture errors, root cause analysis, automatic repair, fix verification, no manual intervention required