An open training platform for ethical AI agents. OpenSkills-compatible skill manifests with built-in guardrails. When an agent learns a skill here, it inherits the safety constraints.