cupynumeric-hdf5

Pass

Audited by Gen Agent Trust Hub on Jun 18, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is authored by NVIDIA Corporation, a trusted entity. All external references, including documentation links (docs.nvidia.com) and repository URLs (github.com/nv-legate), point to legitimate vendor-controlled infrastructure.
  • [EXTERNAL_DOWNLOADS]: The skill documents a dependency on 'h5py', which is a standard library for HDF5 I/O. It recommends installation via 'conda-forge', a well-known and trusted package registry. This is a standard practice for managing scientific Python dependencies.
  • [COMMAND_EXECUTION]: The skill includes shell commands for environment configuration (e.g., 'export LEGATE_IO_USE_VFD_GDS=1') and running verification scripts (e.g., 'legate --io-use-vfd-gds my_script.py'). These commands are standard for the Legate runtime environment and do not involve unauthorized privilege escalation or dangerous operations.
  • [DATA_EXFILTRATION]: The skill facilitates reading and writing HDF5 files. While it involves data movement between memory and disk, these operations are restricted to the user's local or distributed filesystem as part of the intended high-performance computing workflow. No unauthorized network exfiltration patterns were detected.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 18, 2026, 04:34 PM
Security Audit — agent-trust-hub — cupynumeric-hdf5