Contributing

Thank you for improving AGENTS.md Generator. This repository is an agent skill first: changes should help an AI coding agent create, review, and verify coding-agent instruction files more reliably.

Contribution Principles

• Keep SKILL.md concise and operational.
• Move detailed background, command behavior, review policy, and long examples into references/.
• Keep deterministic repository inspection, rendering, verification, audit, and shim logic in scripts/.
• Do not claim a command is verified unless it was actually run.
• Preserve hand-written content outside managed generated blocks.
• Keep generated outputs, temporary profiles, local credentials, and machine-specific paths out of commits.

Suggested Workflow

1. Open an issue describing the agent behavior, stale-doc problem, verification gap, or documentation improvement.
2. Make a focused change with a clear before/after behavior.
3. Run the relevant static validation and skill audit.
4. Include command output or validation evidence in the pull request.

Validation

Useful local commands:

python scripts/inspect_project.py .
python scripts/detect_scopes.py .
python scripts/render_agents.py .
python scripts/verify_agents.py .
python scripts/audit_skill.py .

For script changes, also run a Python syntax check over scripts/*.py.

Documentation Expectations

• Keep the default README.md in English.
• Put Chinese user-facing documentation in README-CN.md.
• Keep examples short, reproducible, and aligned with the skill's AGENTS.md-only scope.