tooluniverse-dataset-discovery
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to use the
requestslibrary to fetch dataset files from external URLs and scientific repositories (e.g., GEO, SRA, ClinicalTrials.gov). While these are reputable scientific sources, the skill facilitates the retrieval of arbitrary data from the web. - [COMMAND_EXECUTION]: The instructions direct the agent to 'Write and run Python code via Bash' to perform data analysis and visualization. This involves executing locally generated scripts that interact with the system environment to process data and save files (e.g., saving plots to
/tmp/). - [INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection because it ingests and processes untrusted data from external sources.
- Ingestion points: Dataset files (CSV, JSON, XPT, etc.) and search result metadata from various genomic and clinical databases.
- Boundary markers: No specific boundary markers or instructions to ignore embedded commands are present in the code snippets that process external data.
- Capability inventory: The skill has access to network operations (
requests), shell execution (bash), and file system writes (plt.savefig). - Sanitization: The skill relies on standard Python parsing libraries (
pandas) which handle data structure but do not sanitize semantic content that might be interpreted by the LLM as instructions if the data is printed to the console.
Audit Metadata