tooluniverse-cancer-genomics-tcga
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes clinical and genomic data from external repositories which could contain malicious instructions.
- Ingestion points: SKILL.md (via tools like GDC_get_clinical_data and direct API calls to api.gdc.cancer.gov).
- Boundary markers: Absent; external data is processed and integrated into the agent's context without explicit delimiters or instructions to ignore embedded content.
- Capability inventory: The skill allows Python code execution via the requests and pandas libraries to fetch, normalize, and report on data.
- Sanitization: Absent; the skill does not specify methods to sanitize or filter incoming biological metadata before it is presented to the user or used in further logic.
- [COMMAND_EXECUTION]: The skill provides Python code templates that perform network requests and data processing when executed by the agent.
- [EXTERNAL_DOWNLOADS]: The skill facilitates the download of genomic data files (such as MAF mutation files) from the official National Cancer Institute Genomic Data Commons (GDC) API.
- [DYNAMIC_EXECUTION]: The skill generates Python code snippets for bulk data retrieval and clinical data wrangling, which are intended to be executed at runtime to process results.
Audit Metadata