A structured logging and promotion system that enables AI coding agents to learn from errors, user corrections, and discovered best practices.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install self-improving-agent-1-0-5
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 self-improving-agent-1-0-5 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 Agent Skill?
The Self-Improvement skill is a powerful framework designed to facilitate continuous learning for AI coding agents. It provides a systematic way to capture unexpected command failures, user corrections, and newly discovered insights directly into markdown files. By integrating this into Openclaw Skills, developers ensure that their agents don't just execute tasks but actually evolve based on the context of each session.
At its core, the skill focuses on turning tactical observations into strategic project memory. Whether it is a misunderstood API or a specific project convention, this skill allows the agent to log the event, assign it a priority, and eventually promote it to permanent files like CLAUDE.md or SOUL.md. This prevents the recurrence of mistakes and streamlines the developer's experience by building a persistent knowledge base.
Self-Improvement Agent Skill Use Cases
- Logging unexpected command or operation failures for later analysis and fixing.
- Recording user corrections when the agent provides incorrect or outdated information.
- Tracking feature requests and missing capabilities identified during a session.
- Documenting newly discovered best practices or workflow improvements for recurring tasks.
- Reviewing past learnings and errors before beginning a major new development task.
How Self-Improvement Agent Skill Works
- The agent identifies a trigger such as a failed bash command, a specific user correction, or a new knowledge discovery.
- A structured entry is generated using a specific ID format (e.g., LRN-YYYYMMDD-XXX) and appended to the relevant file in the .learnings directory.
- The agent assigns metadata to the entry, including priority levels, area tags (like frontend or infra), and a pending status.
- During natural breakpoints, the agent or developer reviews these logs to resolve entries or link them to recurring patterns.
- Broadly applicable learnings are promoted to long-term project memory files to guide future agent behavior and tool usage.
Self-Improvement Agent Skill Setup
The easiest way to get started with this skill is through the recommended installation method for Openclaw Skills:
clawdhub install self-improving-agent
Alternatively, you can manually clone the repository into your skills directory:
git clone https://github.com/peterskoett/self-improving-agent.git ~/.openclaw/skills/self-improving-agent
After installation, initialize the learning directory structure within your workspace:
mkdir -p ~/.openclaw/workspace/.learnings
Self-Improvement Agent Skill Data Schema & Taxonomy
The skill organizes its data within a dedicated .learnings/ directory using a standardized markdown taxonomy:
| File | Content Type |
|---|---|
| LEARNINGS.md | Corrections, knowledge gaps, and technical best practices. |
| ERRORS.md | Command failures, exceptions, and stack traces. |
| FEATURE_REQUESTS.md | New capabilities or tools requested by the user. |
Each entry follows a strict schema including a unique ID, timestamp, priority (low to critical), status (pending, resolved, promoted), and metadata tags for specific codebase areas.
Self-Improvement Agent Skill Advanced Features
- Automatic skill extraction that converts verified, recurring learnings into standalone Openclaw Skills using helper scripts.
- Inter-session communication tools like sessions_send and sessions_history to share learnings across different agent instances.
- Post-tool execution hooks that automatically detect bash errors and prompt the agent to log a failure entry.
- Promotion targets that allow distilled knowledge to be integrated into AGENTS.md, SOUL.md, or TOOLS.md for workspace-wide impact.
- Multi-agent compatibility with specific configurations for Claude Code, Codex CLI, and GitHub Copilot.
SKILL.md
Loading
Related Openclaw Skills
PinchBoard
A specialized social networking protocol for AI agents to communicate, share status updates, and engage with a machine-centric community.
OpenAI Whisper CLI
A powerful command-line tool for local speech-to-text transcription and translation without cloud dependency.
Video Pro
Video Pro is a professional AI video generation solution that transforms text scripts into high-quality short videos using batch processing and diverse templates.
WeWork Archive & Callback Service
A professional Enterprise WeChat integration service providing message event callbacks and secure session content archiving for enterprise compliance.
Daily Game News Skill
An automated scraping and reporting tool that aggregates news from top gaming websites into structured daily digests.
Trakt API Integration
A powerful integration for managing movie and TV show watchlists, history, and ratings via the Trakt.tv API.
Featured*
