audit-context-building
Pass
Audited by Gen Agent Trust Hub on May 11, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill defines a process for granular codebase analysis, focusing on architectural understanding and invariant mapping. It explicitly separates context building from vulnerability identification to ensure accuracy.- [SAFE]: The 'function-analyzer' subagent is configured with standard code-reading tools (Read, Grep, Glob). These are appropriate for the skill's purpose and are not misused for exfiltration or sensitive file access.- [SAFE]: The skill includes extensive anti-hallucination measures, requiring line-number citations for all claims and explicit documentation of assumptions and unknowns.- [SAFE]: No evidence of prompt injection, obfuscation, or remote code execution was found. The instructions prioritize neutral, structural observations over conclusions.
Audit Metadata