openclaw-workspace

Summary

This workspace provides a well-structured environment for building and managing self-improving agent skills for OpenClaw. It includes documentation, templates, scripts, and examples to help you capture learnings, log errors, define tools, and extract reusable skills for AI assistants.

Included Assets

• AGENTS.md
  Guidelines for agent behavior, workspace memory management, and workflow.
• SOUL.md
  Core behavioral principles and boundaries for agents.
• TOOLS.md
  Space to document environment-specific tools and local settings.
• skills/self-improving-agent-3.0.10/
  Self-improvement skill with documentation, scripts, assets, and integration resources:
  • SKILL.md — Description and usage of the self-improvement skill.
  • assets/LEARNINGS.md, ERRORS.md, FEATURE_REQUESTS.md — Logs for knowledge, errors, and requests.
  • assets/SKILL-TEMPLATE.md — Skill documentation template.
  • hooks/openclaw/ — Hook for automatic learning capture.
  • references/ — Examples and integration guides.
  • scripts/activator.sh — Learning capture reminder script.
  • scripts/error-detector.sh — Error detection and logging script.
  • scripts/extract-skill.sh — Helper for extracting reusable skills.
  • _meta.json — Metadata for the skill.
• .openclaw/workspace-state.json
  Workspace state and metadata file.

How to Use

1. Familiarize Yourself

   • Read AGENTS.md for workspace organization.
   • Review SOUL.md to understand agent principles and boundaries.
   • Use TOOLS.md to note down local environment configurations.

2. Self-Improvement Skill Setup

   • See skills/self-improving-agent-3.0.10/SKILL.md for details on capturing and managing learnings.
   • Use .learnings/LEARNINGS.md, ERRORS.md, and FEATURE_REQUESTS.md for structured knowledge capture.
   • Refer to assets/SKILL-TEMPLATE.md to create or extract skills from recurring learnings.

3. Integrating and Automating

   • Use scripts/activator.sh as a prompt reminder after each user instruction.
   • Use scripts/error-detector.sh to detect command failures for logging.
   • Use scripts/extract-skill.sh to generate new skills from learning entries.
   • For hook-based automation, review references/hooks-setup.md and hooks/openclaw/HOOK.md.

4. Practice and Improve

   • Log errors, learnings, and feature requests as they appear in your workflow.
   • Regularly review and update your logs, extracting reusable skills when applicable.

Notes

• The workspace is designed for Shell environments, with scripts and documentation provided in Markdown and Bash.
• Memory files (memory/YYYY-MM-DD.md, MEMORY.md) are referenced in AGENTS.md but may need to be created as you use the workspace.
• Never log sensitive information (e.g., secrets or tokens) into .learnings/ or other shared files.
• Hooks and scripts aim to optimize learning retention and promote best practices for AI code assistance.
• For best results, keep your local notes in TOOLS.md separate from reusable skills and logs.

Refer to the individual documentation files in each directory for more detailed usage instructions and conventions.