skills/nvidia/skills/complexa-sweep/Gen Agent Trust Hub

complexa-sweep

Pass

Audited by Gen Agent Trust Hub on Oct 28, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local shell commands to drive the Proteina-Complexa design pipeline and its utility scripts.
  • The SKILL.md file orchestrates a workflow using complexa design and script_utils/generate_inference_configs.py.
  • The scripts/preflight.sh script probes the local system for GPU, disk, and tool readiness, emitting the results as a JSON file (./complexa_setup/preflight.json).
  • The scripts/write_manifest.py script captures run metadata, including Git SHA and truncated SHA-256 hashes of checkpoint files, to facilitate experiment tracking.
  • [SAFE_PRACTICE]: The scripts/preflight.sh script parses the repository's .env file as inert data to identify required environmental paths (e.g., AF2_DIR, CKPT_PATH) rather than sourcing it directly, which prevents accidental execution of environment values.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 28, 2026, 10:10 PM