agents
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill defines a structured 'Operating contract' that enforces safety boundaries, requiring user confirmation for destructive remote actions and changes outside the local repository. It emphasizes clear success criteria and verification of work.
- [SAFE]: The 'slap' command (emergency stabilization) provides a detailed protocol for draining runaway workflows and containing damage. It specifically instructs the agent to monitor for leaked secrets in logs and prevents the re-injection of sensitive data into the prompt.
- [SAFE]: Subagent orchestration instructions (the 'sub' command) implement best practices for isolation, including parallel exploration for research and sequential writes to prevent overlapping ownership of files. It leverages standard platform tools and official configurations for well-known agent harnesses like Claude Code, Codex, and Cursor.
- [SAFE]: No obfuscation, remote code execution, or unauthorized external downloads were detected. All external references point to official documentation for established technology services.
Audit Metadata