skills/aipoch/open-science/boltz/Gen Agent Trust Hub

boltz

Pass

Audited by Gen Agent Trust Hub on Jul 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill relies on the boltz PyPI package and model weights hosted on the official GitHub repository (github.com/jwohlwend/boltz). These are standard components for the Boltz-2 structure prediction model.
  • [DATA_EXFILTRATION]: When the --use_msa_server flag is enabled, the tool transmits user-provided protein sequences to the ColabFold MSA server at api.colabfold.com (operated by the Steinegger Lab). This is a functional requirement for generating Multiple Sequence Alignments (MSA) necessary for accurate structure prediction, but it involves sending research data to an external service provider.
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute the boltz predict command line interface. This is the intended primary purpose of the skill for performing biomolecular simulations and does not involve suspicious command patterns.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 14, 2026, 10:23 PM
Security Audit — agent-trust-hub — boltz