blueprint

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

• Third-party content exposure detected (high risk: 0.75). The skill’s Phase 1 “Recon + Ingest” workflow reads “pointed-at sources” (docs/issues/URLs/directories) in full via read-only subagents and then consolidates that content into the main agent’s context, which can include outsider-authored free text from public web content or third-party issue/comment bodies.