projects

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

• Third-party content exposure detected (high risk: 0.90). The skill's Phase 1 workflow explicitly instructs the agent to ingest external content—e.g., "Google Doc dump: Extract structure" and "check the codebase, existing reports, and web before accepting the gap" with fallback to "WebSearch for web context" and dispatching /research or /worldmodel—so it fetches and reads untrusted public/user-generated web content that will be interpreted and used to drive decisions.