fiftyone-dataset-import

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

• Third-party content exposure detected (high risk: 0.85). The required workflow includes importing from Hugging Face Hub by scanning/downloading a user-specified repo_id at runtime (e.g., list_repo_files, hf_hub_download, snapshot_download), which ingests outsider-authored README/parquet/COCO/YOLO files into the agent/LLM context as readable text/metadata during the scan and import decision steps.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly performs runtime installs from remote Git repositories which execute fetched code as part of installation (e.g., pip install "git+https://github.com/scaleapi/pandaset-devkit.git#subdirectory=python" and the generic git+https://github.com//.git or git clone https://github.com//.git && pip install . patterns), and those external packages are required for importing certain dataset formats.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)