autonomous-workflow

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

• Third-party content exposure detected (high risk: 0.90). The planner agent is explicitly allowed WebFetch/WebSearch and Phase 1 planning invokes parallel "Explore" sub-agents for research (see CLAUDE.md tool table: "WebFetch / WebSearch | Yes" and SKILL.md/phase-1-planning descriptions), so the workflow may fetch and interpret public web/third‑party content which can influence plan.md and downstream actions.