skills/nvidia/skills/multi-node-slurm/Gen Agent Trust Hub

multi-node-slurm

Pass

Audited by Gen Agent Trust Hub on May 25, 2026

Risk Level: SAFECREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The skill templates instruct users to export sensitive API tokens such as GH_TOKEN, HF_TOKEN, and WANDB_API_KEY as environment variables. These variables are then used within shell commands executed via srun and piped to tee. This workflow creates a risk that credentials could be recorded in log files stored on shared filesystems, potentially exposing them to other users with access to those logs.
  • [PROMPT_INJECTION]: The skill defines a debugging workflow where the agent parses untrusted Slurm logs to diagnose failures. * Ingestion points: Slurm job logs are read using grep patterns in the Debugging Multi-Node Failures section. * Boundary markers: None are specified to separate log content from agent instructions. * Capability inventory: The skill has access to shell execution via srun, file system operations, and network-connected tools. * Sanitization: No sanitization or validation of the log content is performed before interpretation.
Audit Metadata
Risk Level
SAFE
Analyzed
May 25, 2026, 02:31 AM
Security Audit — agent-trust-hub — multi-node-slurm