audit-context-building
Pass
Audited by Gen Agent Trust Hub on May 11, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides structured instructions for code analysis and does not contain any malicious code, prompt injections, or unauthorized data access.\n- [PROMPT_INJECTION]: No attempts to bypass safety filters or override system instructions were found. The instructions are focused on guiding the agent's analytical process for a specific task (code auditing).\n- [EXTERNAL_DOWNLOADS]: No external URLs or remote resource downloads are present in the skill files.\n- [DATA_EXFILTRATION]: There are no commands or instructions that attempt to access sensitive local data or exfiltrate information to external servers.\n- [DATA_EXPOSURE]: The skill analyzes external code provided by the user but includes anti-hallucination and evidence-based verification requirements (e.g., line number citations) to ensure the analysis remains grounded in the provided source code.
Audit Metadata