github

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly clones and pulls arbitrary GitHub repositories via commands like git clone git@github.com:<owner>/<repo>.git (see "Clone third-party repo" and examples such as Clone anthropics/claude-code), thereby ingesting untrusted, user-generated content from the public web into the agent workflow where it will be read/interpreted.