realtime-features

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

• Third-party content exposure detected (high risk: 0.80). The skill's SKILL.md shows the agent subscribing to and handling live user-generated content from external sources—e.g., Supabase realtime channels (.from('messages')/.channel(...) with postgres_changes and broadcast handlers) and an SSE client (useSSE(url))—and it directly reads payloads to update state and trigger actions (setMessages, invalidateQueries, broadcast), so untrusted third‑party content can influence runtime behavior.