Contributing Guidelines

This guide reflects how I run OpenClaw. Contributions are welcome, but the goal is correctness and useful operational detail, not breadth.

Good contributions

• Fixes for upstream OpenClaw changes
• Corrections to inaccurate or outdated information
• Safer config examples
• Clearer automation, security, memory, or access patterns
• Examples that have been tested with a real OpenClaw setup

Not accepted

• “Best model” lists
• Vendor evangelism
• Rewrites that erase the direct tone
• Generic AI advice not specific to OpenClaw
• Public-exposure shortcuts without clear risk notes
• Instructions to blindly install third-party skills

Skills policy

ClawHub links are fine when they help people inspect ideas or source. The preferred pattern is still to rebuild your own local skill from the idea, source, or behavior you actually need.

Do not submit examples that assume unknown third-party skill code is safe to install and run without review.

Attribution and ownership

Some files in this repo are personal runbook guidance. Others are community contributions or source-review notes for external projects.

• Preserve existing source, author, contributor, and date notes.
• Do not rewrite contributed examples as first-person personal setup claims.
• For external projects, include the source URL and a review date.
• For community showcases, say when the commands are historical or need current-doc verification.
• If a change substantially rewrites someone else’s contributed example, explain that in the PR.

Style

• Be specific.
• Use operational language.
• Prefer examples someone can validate.
• Avoid hype.
• Keep personal claims framed as personal claims.
• Assume the reader will run this long-term.

If in doubt, open an issue before submitting a PR.