remote-compute-modal

Pass

Audited by Gen Agent Trust Hub on Aug 1, 2026

Risk Level: SAFE
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill's primary function is to facilitate the execution of arbitrary shell commands and Python code within isolated remote Modal sandboxes. This allows the agent to run complex bioinformatics pipelines on GPU hardware.
  • [EXTERNAL_DOWNLOADS]: The skill automates the download of various scientific tools and model weights from external sources including GitHub, AWS S3, and university-hosted mirrors (e.g., files.ipd.uw.edu). Integrity is ensured through version pinning and SHA256 verification for sensitive assets like the MMseqs2 binary and AlphaFold2 parameters.
  • [COMMAND_EXECUTION]: The provider shim (provider.py) dynamically constructs and executes lifecycle management scripts, such as a PID-1 bash watchdog, within the remote environment to ensure containers are terminated during inactivity.
  • [REMOTE_CODE_EXECUTION]: The provider.py shim utilizes runtime monkey patching of the grpclib and aiohttp libraries to route all SDK network traffic through a local SOCKS5 proxy. This is a security-enhancing feature that ensures remote computations adhere to the host-specified domain allowlists and network restrictions.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 1, 2026, 05:34 AM
Security Audit — agent-trust-hub — remote-compute-modal