metforge-analysis

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is focused on reproducible atmospheric science calculations using industry-standard libraries such as MetPy, xarray, and SciPy.
  • [DATA_EXPOSURE]: The skill contains an explicit security safeguard instructing the agent to 'Never record credentials or secrets' when appending records to logs or provenance files.
  • [COMMAND_EXECUTION]: The skill permits the use of Slurm (sbatch) for distributed computing tasks, which is an expected and standard practice for the scientific workflows described.
  • [PROMPT_INJECTION]: The instructions are clearly defined for scientific tasks and do not contain any patterns attempting to override agent safety or system instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 01:27 PM
Security Audit — agent-trust-hub — metforge-analysis