depmap

Pass

Audited by Gen Agent Trust Hub on Oct 23, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches scientific datasets and genomic metadata from the DepMap portal and Figshare. These are well-known and reputable repositories for scientific research data.
  • Evidence: https://depmap.org/portal/download/all/, https://figshare.com/ndownloader/files/..., https://figshare.com/articles/dataset/DepMap_24Q4_Public/27993966.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external data from APIs and CSV files which could theoretically contain malicious instructions, though the risk is low for this specific scientific use case.
  • Ingestion points: API responses from depmap.org/portal/api and local loading of large CSV files (e.g., CRISPRGeneEffect.csv, sample_info.csv) using pandas.read_csv.
  • Boundary markers: None present in the instructions to delimit external data from agent commands.
  • Capability inventory: The skill includes network access (requests.get), local file writing (open(output_path, 'wb')), and extensive data processing capabilities.
  • Sanitization: No explicit sanitization or validation logic is provided for the content of the downloaded datasets.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 23, 2026, 03:12 AM