product-management

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly describes autonomous web research (e.g., the "Competitive Intelligence (/pm:landscape, /pm:scout [name])” and agents such as research-agent that use WebFetch/WebSearch to pull data from public sites like linear.app, G2, Reddit, Twitter, LinkedIn, etc.) and then uses that untrusted, user-generated third‑party content to compute WINNING scores and create/prioritize GitHub issues, so external pages can materially influence agent decisions.