tool-design
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists entirely of instructional and educational content regarding the design of tools for AI agents. It does not contain any executable scripts or automation logic that could be exploited.
- [PROMPT_INJECTION]: No prompt injection attempts, role-play overrides, or instructions to bypass safety guidelines were found in the text.
- [REMOTE_CODE_EXECUTION]: The skill does not perform any remote code downloads or execution. References to external repositories like GitHub are used for documentation and attribution purposes only.
- [COMMAND_EXECUTION]: While the text mentions Unix utilities (e.g., grep, cat, find, ls) in the context of the 'File System Agent Pattern', these are discussed as design examples rather than instructions for immediate execution.
- [DATA_EXFILTRATION]: No sensitive file access or unauthorized network operations were detected. The skill does not attempt to access credentials or environment variables.
- [OBFUSCATION]: The content is clear and uses no encoding (Base64, hex), zero-width characters, or homoglyph techniques to hide instructions or URLs.
Audit Metadata