discord

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s Discord runtime path uses a generic message tool that includes action: "read" and action: "search" (with a query), which can cause the agent to ingest outsider-authored Discord message text that’s provided via the workspace’s channels (e.g., attacker-authored posts that the agent reads/searches).