validate
Pass
Audited by Gen Agent Trust Hub on Feb 19, 2026
Risk Level: SAFENO_CODEPROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection (LOW): The skill directs the agent to research external platforms like Reddit, Google, and competitor review sites. This creates a surface for indirect prompt injection where malicious content on those platforms could influence the agent's behavior. \n
- Ingestion points: Reddit, forums, Google Trends, Twitter/X, and competitor review sites (G2, Capterra) as specified in the 'Problem Validation' and 'Where to Find People' sections of SKILL.md. \n
- Boundary markers: Absent. There are no instructions to the agent to ignore or delimit potentially malicious instructions found within the researched data. \n
- Capability inventory: The skill implies the use of web search or browsing tools to gather market evidence. \n
- Sanitization: Absent. The skill does not provide methods for filtering or escaping content from external sources before summarization.
Audit Metadata