alphafold2

Pass

Audited by Gen Agent Trust Hub on Jul 2, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references and uses the public MMseqs2 server at api.colabfold.com. This is the documented and intended behavior for ColabFold to avoid requiring massive local databases. While user sequence data is sent to this external service, it is clearly disclosed in both the metadata and the documentation.
  • [DATA_EXFILTRATION]: Sequence data is transmitted to api.colabfold.com as part of the tool's core functionality. The skill provides clear instructions on how to minimize this interaction by using the --msa-only flag and caching results locally.
  • [COMMAND_EXECUTION]: The skill documents the use of the colabfold_batch command-line tool. It also provides technical guidance on using sed to patch environment-specific issues in batch.py when running under gVisor sandboxes to prevent infinite memory allocation loops.
  • [SAFE]: The external resources referenced, such as Google DeepMind's AlphaFold repository and the Steinegger Lab's ColabFold repository, are well-known and reputable sources in the bioinformatics community.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 2, 2026, 01:21 AM
Security Audit — agent-trust-hub — alphafold2