recipe-discover

Pass

Audited by Gen Agent Trust Hub on Jun 16, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements a robust execution protocol with multiple blocking checkpoints ('[STOP — BLOCKING]'). The agent is strictly prohibited from proceeding to sensitive tasks like hypothesis generation or file creation without explicit user confirmation. This ensures human oversight for all synthesized outcomes.
  • [PROMPT_INJECTION]: The skill identifies a potential surface for indirect prompt injection through its ingestion of external data.
  • Ingestion points: User-provided $ARGUMENTS and external data retrieved via the WebSearch tool during market analysis.
  • Boundary markers: The skill does not use specific delimiters to isolate external content.
  • Capability inventory: The agent can invoke sub-agents, perform network searches using WebSearch, and write files to the 'docs/discovery/' directory.
  • Sanitization: No explicit content sanitization or validation logic is defined in the instructions.
  • Mitigation: The risk of indirect injection is effectively neutralized by the requirement for manual user approval before the agent takes any persistent actions based on the processed data.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 16, 2026, 06:05 AM
Security Audit — agent-trust-hub — recipe-discover