A sophisticated logging and optimization framework that enables AI agents to learn from errors, user corrections, and discovered best practices for continuous growth.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install warren-self-improving-agent-v2
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install warren-self-improving-agent-v2 using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Self-Improvement AI Agent Skill?
The Self-Improvement skill is a powerful mechanism designed to facilitate continuous learning and optimization within AI agent environments. By systematically capturing command failures, user corrections, and feature gaps, it allows coding agents to evolve and refine their behavior over time. This skill leverages a structured logging approach within Openclaw Skills to ensure that temporary fixes and non-obvious solutions are transformed into permanent project memory or reusable skills, significantly reducing technical debt and repetitive errors.
Self-Improvement AI Agent Skill Use Cases
- Capturing unexpected command or operation failures for future prevention and troubleshooting.
- Documenting user corrections when the agent makes a mistake or provides outdated information.
- Tracking requested features that are currently missing from the agent's capabilities to prioritize development.
- Identifying recurring patterns in development workflows to automate and harden processes via Openclaw Skills.
- Promoting project-specific conventions to permanent memory files like CLAUDE.md or SOUL.md.
How Self-Improvement AI Agent Skill Works
- The agent detects a trigger such as an error, a user correction, or a newly discovered best practice during a session.
- A structured entry is created in the .learnings directory (LEARNINGS.md, ERRORS.md, or FEATURE_REQUESTS.md) with metadata like priority and technical area.
- The agent or developer reviews logged entries at natural breakpoints or via automated session hooks.
- High-value learnings are promoted to workspace-level memory files or extracted into standalone Openclaw Skills for broader use.
- Recurring patterns are identified through stable deduplication keys, triggering systemic fixes or updated automated guidance.
Self-Improvement AI Agent Skill Setup
Installation is streamlined for developers using the command line.
# Install via ClawdHub (Recommended)
clawdhub install self-improving-agent
# Manual Installation
git clone https://github.com/peterskoett/self-improving-agent.git ~/.openclaw/skills/self-improving-agent
# Initialize the learning directory in your workspace
mkdir -p ~/.openclaw/workspace/.learnings
# Enable automatic evaluation hooks (optional)
openclaw hooks enable self-improvement
Self-Improvement AI Agent Skill Data Schema & Taxonomy
The skill organizes its data within a dedicated .learnings/ directory using standardized Markdown templates to ensure compatibility across Openclaw Skills.
| File | Purpose | Key Metadata |
|---|---|---|
LEARNINGS.md |
Corrections and best practices | Priority, Area, Pattern-Key, Recurrence-Count |
ERRORS.md |
Logged failures and exceptions | Error message, Command context, Reproducibility |
FEATURE_REQUESTS.md |
User-requested capabilities | Complexity, Frequency, Suggested Implementation |
Self-Improvement AI Agent Skill Advanced Features
- Automated skill extraction scripts to convert proven learnings into reusable Openclaw Skills with a single command.
- Inter-session communication tools like sessions_send and sessions_spawn for complex multi-agent coordination.
- Hook integration for real-time error detection and post-task evaluation reminders.
- Recurring pattern detection with automatic promotion rules based on frequency (e.g., 3+ occurrences) and recency.
- Multi-agent support providing consistent learning patterns across Claude Code, Codex, and GitHub Copilot.
SKILL.md
Loading
Related Openclaw Skills
Fake Identity Generator
A specialized AI skill for generating context-aware, realistic fake identities including government IDs, addresses, and integrated temporary email accounts.
Popeye Career Assistant
A comprehensive career advancement tool designed to optimize resumes, predict interview questions, and draft professional work reports.
Mobile ATC
Mobile ATC is a specialized diagnostic skill designed to identify and resolve mobile-specific add-to-cart friction to close the conversion gap between desktop and mobile users.
Summarize PDF
A high-performance PDF-to-Markdown converter that extracts text, tables, and complex formulas using the MinerU Open API.
Saturnday
Saturnday is a governed AI software execution engine that enforces 48+ deterministic security and quality checks on AI-generated code.
Human-Type
A sophisticated keystroke simulation tool that mimics realistic human typing patterns, including errors and deferred corrections, within browser-based environments.
Featured*
