agent-orchestration

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md delegation template explicitly instructs sub-agents to gather and verify information from external "External resources (CI/CD logs, API responses, configurations)" and "Known issues, forums, GitHub issues if relevant" (see the Delegation Template YOUR TASK step 2) and the references (references/accessing_online_resources.md) document using WebFetch/Exa/Ref to read public web pages — so the agent is required to fetch and interpret untrusted, user-generated third‑party content as part of its workflow.