community-builder

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

• Third-party content exposure detected (high risk: 0.90). The skill's required workflow (Daily Operations checklist and Phase 4 — Pipeline Signal Detection) explicitly instructs the agent to monitor and ingest user-generated content from public platforms like Discord, Slack, GitHub, Reddit, and LinkedIn — including capturing verbatim "signal_text" and routing actions based on that content — which exposes it to untrusted third-party inputs that could inject instructions.