fairchem
Pass
Audited by Gen Agent Trust Hub on Jul 2, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill documentation and examples align perfectly with the legitimate use of Meta's FAIRChem library for scientific calculations.
- [EXTERNAL_DOWNLOADS]: The skill instructs users to install standard, well-known packages (
fairchem-core,huggingface-hub) from official registries for their intended purpose. These are documented as requirements for model access and GPU support. - [COMMAND_EXECUTION]: Shell commands provided (
pip install,huggingface-cli login) are standard tasks for setting up the required computational environment and authenticating with Hugging Face to access models. - [PROMPT_INJECTION]: No attempts to override agent behavior, bypass safety filters, or extract system prompts were detected in the instructions.
- [DATA_EXFILTRATION]: No evidence of unauthorized data access, hardcoded credentials, or exfiltration patterns was detected. The authentication step (
huggingface-cli login) is a legitimate requirement for accessing the UMA models.
Audit Metadata