Files
Clément SAILLANT edc35a62b1 feat: Add onboarding scripts and documentation for OpenClaw
- Created install_packages.sh to automate package installation in Ubuntu VM.
- Developed onboarding_simulation.sh for executing onboarding scripts and displaying guides.
- Added openclaw_check.sh to run OpenClaw doctor and help commands.
- Implemented scan_secrets.sh for scanning sensitive patterns in the VM.
- Introduced onboarding README.md and examples.md for contributor guidance.
- Created guide_contributeur.md and guide_vm_sandbox.md for detailed onboarding instructions.
- Added supports_visuels.md with diagrams and video tutorials for better understanding.
- Developed test_openclaw_actions.py for automated testing of OpenClaw actions.
- Created vm_test_report.md template for documenting VM test results.
- Added setup_python_secure.sh for secure Python environment setup.
- Implemented test_openclaw_sanitizer.py for testing sanitization functionality.
2026-02-19 13:16:22 +01:00
..

OpenClaw Integration

This repository includes optional integration points for OpenClaw, a local agent runtime that allows an operator to trigger automations via chat or other humanintheloop interfaces. In this template, OpenClaw is used solely as a viewer and label manager:

  • OpenClaw does not commit or push code. Any actions that modify the repository must be performed via GitHub Agentic Workflows (ghaw) with safeoutputs. This keeps the write surface minimal and auditable.
  • The only allowed actions are adding/removing ai:* labels on issues and pull requests, and posting sanitized status comments. All comments are processed through tools/ai/sanitize_issue.py before being sent.
  • OpenClaw must run in a sandbox or disposable environment with no access to secrets, following the principle of least privilege. Running OpenClaw on the same machine as your source code or CI system is strongly discouraged【57263998884462†L355-L419】.

Refer to docs/security/anti_prompt_injection_policy.md for more details on the defensive measures and threat model.

Workflows & Sécurité OpenClaw

OpenClaw fonctionne strictement en mode observateur: il ne commite ni ne modifie le code source.

Pour contribuer ou intégrer OpenClaw:

Pour toute question, ouvrir une issue ou consulter la FAQ. Consultez le guide onboarding contributeur : onboarding/README.md Consultez aussi le guide pas-à-pas, FAQ et bonnes pratiques : onboarding/guide_contributeur.md Découvrez des exemples dutilisation : onboarding/exemples.md Accédez aux supports visuels, tutoriels vidéo et scripts de test : onboarding/supports_visuels.md Intégrez OpenClaw en local (VM sandboxée) : onboarding/guide_vm_sandbox.md