researching-codebase

Pass

Audited by Gen Agent Trust Hub on Apr 1, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface as it is designed to ingest and process information from untrusted sources, including local codebase files and external web content.
  • Ingestion points: Codebase files located via subagents and web content retrieved via 'WebFetch' or 'web-researcher'.
  • Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions for the processed data.
  • Capability inventory: File system read access through codebase exploration tools, file system write access for documentation (docs/plans/), and network retrieval capabilities.
  • Sanitization: No specific content sanitization or validation logic is defined for the ingested data.
  • [EXTERNAL_DOWNLOADS]: The skill incorporates functionality to fetch external data to support codebase research.
  • Evidence: Uses the 'WebFetch' tool and 'web-researcher' subagent to access documentation or third-party service information.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 1, 2026, 08:46 AM
Security Audit — agent-trust-hub — researching-codebase