variant-analysis

Pass

Audited by Gen Agent Trust Hub on Apr 28, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill instructions involve analyzing content from external codebases, which is a potential surface for indirect prompt injection. However, this risk is inherent to the skill's primary purpose as an auditing tool and is handled via standard agent guardrails.
  • Ingestion points: Source code files from the user's repository.
  • Boundary markers: None explicitly defined to separate code content from instructions.
  • Capability inventory: Execution of search commands (ripgrep) and logic analysis.
  • Sanitization: Not mentioned as part of the search process.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 28, 2026, 01:32 AM