omc-plan
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill enforces a strict execution boundary. It is prohibited from running mutation-oriented shell commands or editing source files. Implementation tasks are delegated to specific execution skills (
ralph,team) only after explicit user approval through a structured UI. - [EXTERNAL_DOWNLOADS]: Fetches changelog and documentation from well-known official sources (Anthropic's GitHub and documentation domains). These operations are used for contextual validation and are not executed as code.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted codebase data and user input. This surface is mitigated by a multi-step verification process involving independent 'Architect' and 'Critic' agents that evaluate plan snapshots, iteration limits, and mandatory human-in-the-loop approval before any generated plan is acted upon.
- [SAFE]: No malicious patterns, obfuscation, or unauthorized data access were detected. The state management and handoff policies follow security best practices for AI agent autonomy, including a dedicated stop hook cleanup routine.
Audit Metadata