agency-workflow-patterns

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs the orchestrator to "Fetch Sprint Issues" and to use agents (see SKILL.md) and multiple agents in references/agent-roster.md list the WebFetch tool and social-media roles (e.g., Trend Researcher, Twitter Engager, Reddit Community Builder), indicating it will fetch and read untrusted, user-generated third‑party content (issues, web/social pages) and act on it as part of workflows.