lens

Pass

Audited by Gen Agent Trust Hub on Sep 8, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill operates by ingesting and analyzing external codebase data, which presents a theoretical surface for indirect prompt injection if the analyzed files contain malicious instructions.
  • Ingestion points: The skill reads file contents using multiple search layers (Structure, Keyword, LSP, and Grep) as described in 'reference/search-strategies.md'.
  • Boundary markers: The skill does not currently define specific delimiters or instructions to ignore embedded prompts within the content it analyzes.
  • Capability inventory: The skill possesses the capability to execute system commands such as 'grep' and 'git', and to invoke third-party analysis tools like 'eslint' and 'madge'.
  • Sanitization: The workflow mitigates risks through a rigorous 'Core Contract' that requires every claim to be backed by specific file/line evidence and reported with explicit confidence levels, rather than through automated input sanitization.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 8, 2026, 06:57 AM
Security Audit — agent-trust-hub — lens