mat-lammps-md
Warn
Audited by Socket on Jun 19, 2026
1 alert found:
AnomalyAnomalyexamples/mace/run_mace_na2si3o7_quench.sh
LOWAnomalyLOW
examples/mace/run_mace_na2si3o7_quench.sh
This module appears to be a legitimate workflow orchestrator for structure generation, model loading/compilation, and LAMMPS simulation. However, it carries a high-impact supply-chain risk: it may download and then deserialize PyTorch checkpoints via torch.load without integrity verification, which can enable arbitrary code execution if an attacker controls or tampers with the checkpoint/model artifact. Additionally, it executes an externally specified LMP_BIN from environment variables, which can increase risk in compromised CI/runtime environments. No overt malware behaviors (exfiltration/backdoor/credential theft) are evident in the provided snippet beyond these trust-boundary issues.
Confidence: 62%Severity: 66%
Audit Metadata