fiftyone-dataset-import

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly downloads and scans arbitrary Hugging Face repositories (HF-HUB-IMPORT.md: list_repo_files, snapshot_download, load_from_hub) and instructs web/GitHub/PyPI searches in SKILL.md to discover installation instructions and import code, meaning untrusted, user-generated third-party content is fetched and parsed and can change import/installation actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs installing packages from GitHub at runtime (which fetches and executes remote code), e.g. pip install "git+https://github.com/scaleapi/pandaset-devkit.git#subdirectory=python", making that URL a runtime external dependency that can execute remote code.