uv-advanced-dependencies

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

• Third-party content exposure detected (high risk: 0.80). SKILL.md and REFERENCE.md explicitly instruct fetching and installing from arbitrary public sources (e.g., "uv add git+https://...", "uv add https://...", tool.uv.extra-index-url and find-links like https://download.pytorch.org/whl/...), meaning the agent will ingest untrusted third‑party repository/package content that can influence installs and subsequent actions.