pm-workflow

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

• Third-party content exposure detected (high risk: 0.80). The workflow explicitly instructs the PM agent to "conduct analysis research (use the internet, mcps, skills)" when a task is blocked and to run "skill research" (Phase 8) to update skills, which indicates the agent will fetch and interpret internet/public third‑party content as part of its decision-making (SKILL.md failure handling and Phase 8).