flows-agent

Pass

Audited by Gen Agent Trust Hub on Apr 22, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill inherently processes untrusted data from the internet as its primary function. It allows the agent to fetch content from over 1000 external API providers through the call_paid_api tool.
  • Ingestion points: Untrusted content is ingested from remote servers via the call_paid_api tool in SKILL.md.
  • Boundary markers: No specific delimiters or safety wrappers are defined for the returned API responses.
  • Capability inventory: The skill provides tools for searching, pricing, and executing remote API calls, but does not grant direct file system or shell access.
  • Sanitization: There is no explicit sanitization or filtering of API responses mentioned in the instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 22, 2026, 12:43 AM