← Back to Skills Marketplace
Self Improving Agent
by
WuXian-upup
· GitHub ↗
· v1.0.0
· MIT-0
204
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install self-improving-agent-zh
Description
Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Clau...
Usage Guidance
This skill appears coherent and focused on logging learnings. Before installing: (1) Inspect the scripts (activator.sh, error-detector.sh, extract-skill.sh) and hook handlers to confirm they contain only the benign behavior you expect (they do in this package). (2) If you install hooks, be aware they run with the agent's user permissions and will inject reminders into every session — enable them only if you want that persistent behavior. (3) When using extract-skill.sh, prefer --dry-run first and avoid running scripts as root. (4) If you don't use OpenClaw hooks, you can still use this skill manually by adding .learnings/ files without enabling hooks.
Capability Analysis
Type: OpenClaw Skill
Name: self-improving-agent-zh
Version: 1.0.0
The 'self-improving-agent' skill bundle is a productivity framework designed to help AI agents log errors, user corrections, and new patterns into markdown files (e.g., `.learnings/`, `SOUL.md`, `AGENTS.md`). It includes utility scripts like `extract-skill.sh` for scaffolding new skills and OpenClaw hooks (`handler.js`) that inject reminders into the agent's bootstrap process. While the skill encourages the agent to modify its own behavioral guidelines and workspace files, the logic is entirely local, lacks any network exfiltration or obfuscation, and serves the stated purpose of session-to-session memory and optimization.
Capability Assessment
Purpose & Capability
The skill's name/description (capture learnings, errors, promote to workspace) matches the provided files and scripts: activator and error-detector reminders, an extractor for creating new skills, and an OpenClaw hook that injects a reminder at bootstrap. Required credentials/env/config are none, which is reasonable for this local logging workflow.
Instruction Scope
Runtime instructions tell the agent and user to create .learnings/, optionally copy/enabled hooks under ~/.openclaw/hooks, and to use scripts that read the CLAUDE_TOOL_OUTPUT env var. The instructions do ask to copy files into the user's OpenClaw workspace and enable hooks (which will affect session bootstrap and PostToolUse events) — this is coherent with the stated purpose but does give the skill a persistent hook into agent sessions when enabled.
Install Mechanism
No automated install spec is included. The README suggests cloning from GitHub or using ClawdHub; scripts are provided locally. There is no download-from-arbitrary-URL or opaque installer — manual git clone is the only remote fetch mentioned. Scripts are included as plain shell/JS files and would be written to disk only if the user installs them.
Credentials
The skill declares no required env vars or credentials. The error-detector script reads CLAUDE_TOOL_OUTPUT (an agent-provided env var) which is expected for PostToolUse hooks; no unrelated secrets or cloud credentials are requested.
Persistence & Privilege
The skill expects optional installation of an OpenClaw hook that injects a virtual reminder at agent bootstrap and a PostToolUse hook that reads tool output. This is appropriate for a self-improvement reminder but does create persistent behavior in the user's OpenClaw environment if they enable the hooks. always:true is not set.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install self-improving-agent-zh - After installation, invoke the skill by name or use
/self-improving-agent-zh - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Self-Improvement Skill v1.0.0
- Initial release of the self-improvement skill for agents.
- Provides structured guidelines for logging errors, learnings, and feature requests to markdown files.
- Includes procedures for promoting important learnings to persistent project memory files (e.g., CLAUDE.md, AGENTS.md, TOOLS.md, SOUL.md).
- Offers quick references for when and how to log various improvement events.
- Contains detailed workflow and markdown templates for learnings, errors, and feature requests.
- Designed for integration with OpenClaw, but also supports generic agent setups.
Metadata
Frequently Asked Questions
What is Self Improving Agent?
Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Clau... It is an AI Agent Skill for Claude Code / OpenClaw, with 204 downloads so far.
How do I install Self Improving Agent?
Run "/install self-improving-agent-zh" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Self Improving Agent free?
Yes, Self Improving Agent is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Self Improving Agent support?
Self Improving Agent is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Self Improving Agent?
It is built and maintained by WuXian-upup (@wuxian-upup); the current version is v1.0.0.
More Skills