utility-pm-workflow-orchestrator

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

• Third-party content exposure detected (medium risk: 0.30). In the non-Claude client runtime path, the workflow executes a user-provided Mode A/B plan/chain by reading each downstream skill’s SKILL.md and then running the engine body inline in the same context window, which means outsider free text from the user’s input is consumed for step parsing and run control (“build the ordered step list (Mode A parse or Mode B named chain)”), i.e., the workflow ingests untrusted text before selecting a specific item.