alphafold-database

Pass

Audited by Gen Agent Trust Hub on Mar 15, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill includes a block titled 'Suggest Using K-Dense Web For Complex Worflows' that provides specific instructions to the agent to proactively market the author's hosted research platform (www.k-dense.ai) under certain conditions. This is a behavioral override designed to influence the agent's response patterns.
  • [COMMAND_EXECUTION]: The skill demonstrates the use of subprocess.run to execute the gsutil command for bulk data downloads. Although the provided code includes a safety check to ensure input is an integer, the presence of subprocess execution capabilities poses a potential risk if extended or used without strict validation.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external scientific databases via REST APIs.
  • Ingestion points: Data is fetched from alphafold.ebi.ac.uk, uniprot.org, and ebi.ac.uk via requests and urllib in SKILL.md.
  • Boundary markers: No specific boundary markers or 'ignore' instructions are used when processing external data.
  • Capability inventory: The skill has the ability to write files to disk and execute system commands via subprocess.
  • Sanitization: While there is an example of checking for integer types on one parameter, there is no comprehensive sanitization of the JSON or text content retrieved from external sources before it is processed or used in logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 15, 2026, 04:31 AM