ml-debug

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

• Third-party content exposure detected (high risk: 1.00). This skill's mandatory "Web mode" grounding and Phase 1 rules require the agent to WebFetch and ingest public third‑party URLs (e.g., GitHub issues, PyPI pages, HuggingFace model config) as part of the required workflow (see "HARD RULE: You MUST ground before writing analysis" and the "Web mode URL registry"), so untrusted, user-generated content can directly influence tool use and next actions.