hugging-face-datasets

Pass

Audited by Gen Agent Trust Hub on Jun 20, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill uses official libraries (huggingface_hub, duckdb, datasets) from well-known providers. Credential management is performed via environment variables (HF_TOKEN) rather than hardcoded secrets, which is consistent with secure development practices.- [COMMAND_EXECUTION]: The skill utilizes Python scripts (dataset_manager.py, sql_manager.py) to perform its primary functions. These scripts are invoked via uv run, allowing the agent to automate dataset creation and SQL-based transformations.- [PROMPT_INJECTION]: The skill processes content from external Hugging Face datasets, which represents a surface for indirect prompt injection.
  • Ingestion points: Data is retrieved from external Hugging Face repositories in scripts/sql_manager.py and processed during row additions in scripts/dataset_manager.py.
  • Boundary markers: None identified. The instructions do not include specific delimiters to prevent the agent from interpreting data content as instructions.
  • Capability inventory: Repository management (creation, deletion), file uploads via HfApi, and SQL query execution through DuckDB.
  • Sanitization: Data is processed as structured objects without specific sanitization of the values against embedded natural language instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 20, 2026, 09:25 PM
Security Audit — agent-trust-hub — hugging-face-datasets