real-time-collaboration-engine

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

• Third-party content exposure detected (high risk: 0.80). The skill's runtime patterns clearly ingest and render untrusted, user-generated content from external peers—e.g., Socket.IO/WebsocketProvider messages (socket.on('operation', ...), socket.on('cursor-update', ...)), presence states and avatars rendered via , and awareness.getStates() from Yjs—which means the agent/app reads and interprets arbitrary third‑party inputs as part of its workflow.