weak-signal-synthesizer

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs the agent to monitor and ingest content from public, user-generated and third-party sources—e.g., "niche Reddit communities," academic preprints, GitHub trending, patents, funding and regulatory sites—which the agent is expected to read and interpret, exposing it to untrusted content that could enable indirect prompt injection.