corroboration-skill-builder
Pass
Audited by Gen Agent Trust Hub on May 8, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: No evidence of prompt injection or instructions to bypass safety guidelines. The prompt uses standard variable interpolation for educational content generation.
- [DATA_EXFILTRATION]: No evidence of data exposure or exfiltration. There are no network requests (curl, wget, etc.) or access to sensitive local file paths like SSH keys or environment files.
- [OBFUSCATION]: No obfuscated content detected. The text is plain markdown, and there are no hidden Base64 strings, zero-width characters, or homoglyphs.
- [REMOTE_CODE_EXECUTION]: No remote code execution patterns or unverified dependencies. The skill does not install packages or execute scripts from external URLs.
- [COMMAND_EXECUTION]: No shell command execution or privilege escalation attempts detected. The skill is entirely text-based.
- [DYNAMIC_CONTEXT_INJECTION]: No use of dynamic context injection (the !command syntax) was found in the SKILL.md file.
Audit Metadata