icp-definer

Pass

Audited by Gen Agent Trust Hub on Jul 20, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill's behavior is consistent with its stated purpose of market research and ICP definition.
  • [EXTERNAL_DOWNLOADS]: The skill performs web searches and fetches content from public forums such as Reddit, Hacker News, and industry blogs to identify market trends and customer pain points.
  • [PROMPT_INJECTION]: The skill processes untrusted data from external websites (ingestion points: Reddit, HN, Twitter, and various blogs as specified in SKILL.md). While explicit boundary markers and sanitization are absent, the risk of indirect prompt injection is inherent to web-based research tasks and is mitigated by the use of structured Markdown templates for the synthesis and final report.
  • [DATA_EXFILTRATION]: The skill reads sales context from a dedicated ".sales/" directory to ground its research; this user-provided data is used to inform its web queries to find relevant market segments.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 20, 2026, 06:10 PM
Security Audit — agent-trust-hub — icp-definer