Skip to content

AI Covenant#

This repository is following the AI Covenant.
It defines how AI tools (e.g. chat assistants, coding agents, AI review bots) may be used when contributing.

The core principle is: You own your contributions, regardless of what tools helped create them.
The main rules are:

  • Human in the loop: AI tools must not commit, push or post in issues, PRs or other channels autonomously.
  • No AI authorship: Commit under your own name, never under an AI tool or bot account. No Co-authored-by:, Assisted-by: or similar trailers and no AI markers in commit messages.
  • Disclosure: Routine AI use needs no disclosure, but AI-proposed changes you do not fully understand must be disclosed in the PR description.
  • Good first issues: Issues labelled other: good first issue 🌱 are reserved for newcomers and must not be solved with AI tools.
  • Confidential data: Secrets, personal data, unpublished results and other confidential data must not be shared with AI tools.
  • AI-found bugs: Bugs and vulnerabilities found with AI tools must be reproduced by a human before they are reported, vulnerabilities according to the Security Policy.

Guidance for AI agents#

The 📝 AGENTS.md file provides guidance for AI coding agents working in the repository, including the commands, conventions and the rules of the AI Covenant.
📝 CLAUDE.md is a symlink to 📝 AGENTS.md, so that Claude Code uses the same guidance.

Agent skills#

Skills for AI agents can be added to the repository in 📝 .agents/skills/<skill-name>/SKILL.md.
To make a skill available to Claude Code, add a symlink in 📝 .claude/skills/:

💻 ln -s ../../.agents/skills/<skill-name> .claude/skills/<skill-name>

Symlinks on Windows

Git checks out symlinks as plain text files on Windows, unless symlinks are enabled (Developer Mode and 💻 git config core.symlinks true).

Used Icons

🐙 GitHub | 💠 git | 📝 File | 💻 Command Line