pm-ai-research

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

• Third-party content exposure detected (high risk: 0.90). The skill's SKILL.md explicitly requires ingesting open web and community sources (e.g., Phase 3 "Search for academic research or industry papers" and "Continuous Monitoring Mode" that pulls competitor changelogs, press releases, job postings, analyst coverage and customer community signals like G2, Reddit, Hacker News), and those signals are used to drive decisions and automated actions, which exposes the agent to untrusted third‑party content that could enable indirect prompt injection.