Thank you for your interest in contributing to OpenClaw Agents! 🦞
- Use GitHub Issues to report bugs or request features
- Include steps to reproduce, expected behavior, and actual results
- Label issues appropriately (
bug,enhancement,agent-config, etc.)
- Fork this repository
- Create a branch:
git checkout -b agent/your-agent-name - Create agent files following the Agent Structure Guide
- Update
agents.yamlmanifest - Submit a Pull Request
Each agent requires three core files in .agents/<agent_name>/:
| File | Purpose |
|---|---|
soul.md |
Agent personality, capabilities, and decision-making principles |
agent.md |
Technical configuration, tool permissions, model preferences |
user.md |
Context about the user/team this agent serves |
- Keep changes backward-compatible when possible
- Document why changes are needed in the PR
- Test with
openclaw agents list --bindingsafter changes
- Create workflow file in
.agents/workflows/ - Include YAML frontmatter with
description - Document all phases, roles, and expected outputs
- Reference agents by their standard IDs
- Be respectful and constructive
- Focus on the shared goal of building better AI workflows
- Provide actionable feedback in reviews
- Use Markdown for all documentation
- Keep agent names lowercase with hyphens (e.g.,
data-engineer) - Use emoji consistently for agent identification
- Write documentation in Chinese (primary) with English technical terms
Open a GitHub Discussion or contact the maintainers.