cogworks-encode

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

• Third-party content exposure detected (high risk: 0.90). This skill explicitly ingests and requires reading public third‑party sources (see metadata.json "sources" pointing to https://github.com/williamhallatt/cogworks and the reference.md Input Format showing "Type: {url|file}" and instructions to "read every source completely"), so untrusted, user-generated web content can influence decision rules and downstream actions.