container-management:gpu-bottleneck-analysis
Installation
SKILL.md
GPU Pool Bottleneck Analysis
Analyze GPU cluster capacity across multiple clusters, predict bottlenecks under increased inference QPS, and recommend concrete actions (scale, throttle, or reroute).
REQUIRED SUB-SKILL: Use dce for all command execution, auth checks, and
catalog discovery.
SCOPE: Read-only analysis. This skill does NOT create, modify, or delete any resources.
When to Use
- "If traffic increases by 30%, which GPU pool will bottleneck first?"
- "Should we scale, throttle, or change routing for GPU workloads?"
- "GPU capacity planning across clusters"
- "Which cluster has the most GPU headroom?"
- "Predict GPU saturation under load increase"