agent-code-guide
Pass
Audited by Gen Agent Trust Hub on Feb 21, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [Prompt Injection] (SAFE): The text includes the phrase "forget previous instructions" within the debugging sub-skill. Analysis confirms this is part of a troubleshooting guide for human users to help reset the agent's context during manual sessions, not a malicious payload designed to bypass safety filters or hijack the agent automatically.
- [No Executable Code] (SAFE): The entire skill package consists of Markdown files. There are no Python scripts, Node.js packages, shell scripts, or binaries included that could perform operations on the host system.
- [Data Exposure] (SAFE): No sensitive file paths, credential patterns, or exfiltration vectors were identified in the documentation or templates.
- [External Downloads] (SAFE): The skill contains a reference URL to a GitHub repository for documentation purposes but lacks any mechanisms for automated remote code execution or dependency fetching.
Audit Metadata