← Back to Skills Marketplace
Self Improvement
by
biocrfhkust-cloud
· GitHub ↗
· v0.1.0
· MIT-0
215
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install self-improving-agent-2
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 to do what it claims, but review a few things before enabling hooks: 1) Inspect scripts (activator.sh, error-detector.sh, extract-skill.sh) locally to confirm you’re comfortable with their behavior and make them executable only if appropriate. 2) Be aware CLAUDE_TOOL_OUTPUT may contain command output (potentially secrets); the skill only scans for error patterns but your .learnings/ files may end up containing snippets of tool output—avoid logging sensitive data. 3) When enabling the OpenClaw/agent hook, enable it consciously (it will inject reminders into sessions you opt into). 4) If you plan to run extract-skill.sh, run with --dry-run first and ensure the output directory is within the workspace you intend (the script enforces relative paths). 5) As with any third-party repo, consider vetting the GitHub source and commit history before cloning into ~/.openclaw/skills.
Capability Analysis
Type: OpenClaw Skill
Name: self-improving-agent-2
Version: 0.1.0
The skill bundle is a comprehensive framework for agent self-improvement, allowing an AI to log errors, learnings, and feature requests to markdown files. It includes scripts for error detection (error-detector.sh) and automated skill creation (extract-skill.sh) with proper path validation and naming restrictions to prevent directory traversal. The instructions guide the agent to maintain a persistent memory by updating workspace files like SOUL.md and AGENTS.md, which is consistent with the stated purpose of continuous learning within the OpenClaw ecosystem and lacks any indicators of malicious intent or data exfiltration.
Capability Assessment
Purpose & Capability
Name/description (capture learnings, errors, promote to workspace files) align with what is included: logging templates, hook handlers that inject reminders, an activator and error-detector script, and a helper to extract skills. Nothing in the bundle requires unrelated capabilities such as cloud credentials or network access.
Instruction Scope
SKILL.md and the hooks/scripts restrict behavior to: injecting a virtual reminder at bootstrap, printing lightweight reminders on prompt/command lifecycle events, scanning tool output for error-like patterns, and scaffolding skill files locally. Scripts read CLAUDE_TOOL_OUTPUT (expected platform variable) and create files under relative paths; they do not transmit data externally or access unrelated system config.
Install Mechanism
No automated remote installer or download-from-arbitrary-URL is used. Installation is manual (git clone or clawdhub). All included scripts are present in the repo; no extract-from-unknown-URL behavior was found.
Credentials
The skill declares no required env vars or credentials. The scripts rely on platform-provided CLAUDE_TOOL_OUTPUT to detect errors, which is appropriate for an error-detection hook. No unrelated secret or credential access is requested.
Persistence & Privilege
The skill is not always-enabled and requires opt-in hook configuration. The hook/handler injects a virtual bootstrap file when enabled but does not modify other skills' configs or request elevated/system-wide privileges.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install self-improving-agent-2 - After installation, invoke the skill by name or use
/self-improving-agent-2 - Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.0
Initial release of the self-improving-agent-2 skill, enabling systematic logging and promotion of learnings, errors, and feature requests:
- Provides Markdown-based templates and workflow for continuous learning and self-improvement.
- Organizes logs into `.learnings/LEARNINGS.md`, `ERRORS.md`, and `FEATURE_REQUESTS.md`.
- Documents when and how to log errors, corrections, missing features, outdated knowledge, and best practices.
- Outlines promotion of important learnings to project-wide memory files including `CLAUDE.md`, `AGENTS.md`, and `SOUL.md`.
- Includes setup and integration steps for both OpenClaw and generic agent environments.
- Details structured entry formats, ID conventions, and resolution workflow for effective tracking and improvement.
Metadata
Frequently Asked Questions
What is Self Improvement?
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 215 downloads so far.
How do I install Self Improvement?
Run "/install self-improving-agent-2" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Self Improvement free?
Yes, Self Improvement is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Self Improvement support?
Self Improvement is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Self Improvement?
It is built and maintained by biocrfhkust-cloud (@biocrfhkust-cloud); the current version is v0.1.0.
More Skills