swarm-expert

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

• Third-party content exposure detected (high risk: 0.70). The skill explicitly ingests and surfaces user-generated content from the swarm (e.g., read-messages and list-channels for chat, and get-tasks/get-task-details which return task title/description/output/logs), so it can read and act on arbitrary untrusted content provided by other users/agents.