coreweave-fabric-diagnostics

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is a specialized diagnostic utility for CoreWeave GPU infrastructure. It uses a bundled Python script (scripts/fabric-check.py) to provide deterministic analysis of network transport logs, avoiding potential LLM hallucinations on critical performance data.
  • [SAFE]: The allowed-tools configuration uses fine-grained scoping for Bash commands, limiting access to kubectl get and python3, which aligns with the principle of least privilege.
  • [SAFE]: All external references and documentation point to trusted or well-known organizations including NVIDIA, CoreWeave, and official GitHub repositories.
  • [SAFE]: The skill maintains a read-only posture. While manual remediation steps involving sudo are documented for the user, the skill itself does not have permissions to execute these commands or modify the cluster state.
  • [SAFE]: Data processing is limited to analyzing user-provided logs and pod specifications locally. No network exfiltration or credential harvesting patterns were detected.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 03:45 AM
Security Audit — agent-trust-hub — coreweave-fabric-diagnostics