How do I trigger the diagnostic collection process?

You can trigger the process within Claude Code using natural language phrases like 'coreweave debug', 'coreweave support', or 'collect coreweave logs'.

Does this skill require a CoreWeave API key?

Yes, it utilizes the COREWEAVE_API_KEY environment variable to verify API connectivity and check rate limit headers.

What tools must be installed for this skill to function?

The skill requires kubectl, tar, curl, and grep to be accessible in the environment to perform the full suite of diagnostic tasks.

What data does the CoreWeave Debug Bundle collect?

The skill collects Kubernetes node status, event logs, GPU allocation details, NVIDIA device plugin health, API connectivity metrics, and logs from failed pods into a single diagnostic archive.

When should I use this diagnostic skill?

Use it when troubleshooting GPU scheduling issues, pod failures like OOMKilled, or when preparing a detailed technical report for CoreWeave's support team.

CoreWeave Cluster Diagnostics

Name: CoreWeave Cluster Diagnostics
Author: jeremylongshore

byjeremylongshore

•

2,083

•

Infraestructura en la Nube

Automates the collection of GPU node health, Kubernetes logs, and API diagnostics for CoreWeave cloud infrastructure troubleshooting.

The CoreWeave Debug Bundle is a specialized Claude Code skill designed to streamline the troubleshooting process for GPU-intensive workloads on CoreWeave infrastructure. It automatically gathers critical diagnostic data—including GPU allocation metrics, NVIDIA device plugin status, failed pod logs, and API connectivity states—into a portable archive for support tickets or internal analysis. This tool is indispensable for DevOps engineers managing inference workloads who need to quickly diagnose why pods are stuck pending, why nodes are underperforming, or why API requests are failing.

Características Principales

01Comprehensive Kubernetes pod status and cluster-level event logging

02Automated diagnostic archive generation (tar.gz) for support tickets

03Real-time GPU resource allocation and device plugin health checks

04CoreWeave API connectivity testing and rate limit header analysis

05Targeted extraction of logs from failed pods and OOMKilled events

062,083 GitHub stars

Casos de Uso

01Preparing comprehensive technical data for CoreWeave support escalations

02Debugging Out-Of-Memory (OOM) failures in large-scale AI inference workloads

03Troubleshooting GPU pods stuck in a 'Pending' state due to quota or allocation issues

Características Principales

01Comprehensive Kubernetes pod status and cluster-level event logging

02Automated diagnostic archive generation (tar.gz) for support tickets

03Real-time GPU resource allocation and device plugin health checks

04CoreWeave API connectivity testing and rate limit header analysis

05Targeted extraction of logs from failed pods and OOMKilled events

062,083 GitHub stars

Casos de Uso

01Preparing comprehensive technical data for CoreWeave support escalations

02Debugging Out-Of-Memory (OOM) failures in large-scale AI inference workloads

03Troubleshooting GPU pods stuck in a 'Pending' state due to quota or allocation issues