product-feasibility-scout
Pass
Audited by Gen Agent Trust Hub on Aug 4, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill instructs the agent to research external websites, documentation, and competitor platforms. This creates a surface for indirect prompt injection, where malicious instructions embedded in external content could be parsed and followed by the agent.
- Ingestion points: Web content from external APIs, documentation sites, and competitor pages processed by the
general-purposesub-agent. - Boundary markers: Absent; the research templates do not include instructions to ignore or delimit potentially untrusted commands found in the source material.
- Capability inventory: The skill launches a
general-purposetask agent, which typically possesses tool access (though the skill specifies "RESEARCH ONLY"). - Sanitization: No sanitization or filtering of external data is specified before the content is processed for the final report.
Audit Metadata