audit-security

Pass

Audited by Gen Agent Trust Hub on Oct 24, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: Potential vulnerability surface in external data ingestion.
  • Ingestion points: Untrusted data enters the agent context through firecrawl_search, firecrawl_scrape, and sentry:search_issues.
  • Boundary markers: The instructions do not define specific delimiters to separate external content from internal logic.
  • Capability inventory: Capabilities include reading and writing local files, searching the web, and accessing Sentry APIs.
  • Sanitization: No explicit sanitization of external content is mentioned.
  • [EXTERNAL_DOWNLOADS]: Fetches security guidelines and vulnerability information from well-known technology organizations and official registries.
  • [COMMAND_EXECUTION]: Employs standard package manager audit tools such as npm audit and pip-audit to identify vulnerable dependencies in the codebase.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 24, 2026, 01:29 PM