skill

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

• Third-party content exposure detected (high risk: 1.00). The /skill setup "Import skill" option explicitly accepts a URL (e.g., GitHub gist) to download and save skill markdown into user/project skill directories, and the doc states Claude will automatically detect and apply skills, so arbitrary third‑party content can be ingested and change agent behavior.