autonomous-agent-patterns
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is an educational resource providing design patterns for building AI agents and does not contain any executable malicious payloads, obfuscation, or injection attempts.
- [SAFE]: A dedicated 'Authorization and implementation limits' section in SKILL.md explicitly warns about the security risks inherent in the conceptual code fragments, specifically highlighting risks associated with tool authorization, filesystem access, and command execution.
- [SAFE]: Potentially dangerous code examples, such as the 'SandboxedExecution' class or 'MCPAgent' tool creation, include 'NotImplementedError' blocks and comments stating that real isolation and validation must be implemented before actual use.
- [SAFE]: The skill documentation includes a security checklist and emphasizes best practices like human-in-the-loop approvals for dangerous operations (e.g., sudo commands) and using limited environment variables for shell execution.
Audit Metadata