mcpField
A
aks-gpu-inference

aks-gpu-inference

github.com/microsoftCloud & DevOps1.6k93 installsUpdated today
Repository

About aks-gpu-inference

aks-gpu-inference is a skill for diagnosing AKS GPU and KAITO issues, including scheduling, VRAM/OOM, readiness, and scaling. It binds investigation to a specific target, separates device-allocation failures from container or host OOM, and reports evidence gaps and confidence.

Key features

  1. 01Target-bound AKS investigation
  2. 02Event, status, and gpuProfile collection
  3. 03GPU memory failure analysis
  4. 04KAITO readiness checks
  5. 05Evidence and confidence reporting

Use cases

  1. 01Investigating GPU workload scheduling failures
  2. 02Distinguishing VRAM allocation issues from OOMKilled
  3. 03Checking why a KAITO workspace is not ready
  4. 04Reviewing GPU workload scaling problems

MCP support

The skill can use a host-advertised Azure read when its schema provides the needed evidence, or use its read-only query references. It requires host authorization for command execution and does not route kubectl through Azure MCP.

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/microsoft/azure-skills --skill aks-gpu-inference

After installing, type/aks-gpu-inferencein Claude Code to confirm it loaded.

Bundled files

References and scripts installed along with SKILL.md.

references/
gpu-cost-and-scaling.md
gpu-observability.md
gpu-scheduling.md
kaito-workspaces.md

Details

From the project README

AKS GPU Inference Day-2

Quick Reference

PropertyValue
Best forAKS GPU/KAITO failures
EvidenceBound target, events/status, gpuProfile, DCGM

When to Use This Skill

Use for scheduling, VRAM/OOM, KAITO readiness, and workload scaling. Route exclusions as described above.

MCP Tools

Use a fitting host-advertised Azure read or the references' read-only queries.

Host Capability Gate

Before executing any referenced pipeline, verify that the host authorizes the required shell and kubectl/az commands against the bound target. File-backed collection also requires approved artifact storage and access to any bundled files it uses. A governed Azure CLI tool alone does not establish these capabilities. Equivalent host reads may replace commands only where their advertised schemas provide the required evidence.

If execution is unavailable or prohibited, say so and analyze supplied or redacted events, status, logs, and metrics, or give the operator a scoped collection plan. State which reads did not run and leave conclusions requiring missing evidence unconfirmed. Never route kubectl through Azure MCP or bypass host policy. The authorization requirements below still apply.

Workflow

  1. Bind subscription, cluster, kube context, pool, and affected resource.
  2. Capture exact events/status and observed gpuProfile; state which reads ran.
  3. Route to scheduling, observability, KAITO, or scaling.
  4. Separate container/host OOM from device-allocation failures. For OOMKilled, correlate container memory limits/usage and node conditions. For Managed+Install device-memory evidence, use the exporter on port 19400 and incident-window FB_USED/FB_FREE; sizing tables do not establish a cause.
  5. For KAITO not-ready, warn that Workspace deletion leaves its GPU pools; cleanup is separate and needs explicit authorization.
  6. Report evidence, confidence, missing evidence, and owner handoff.

Require explicit authorization before scaling, cordon/drain, deletion, add-on enablement, or monitoring mutation.

Error Handling

ConditionResponse
Missing/contradictory evidenceMark unconfirmed; request the owner/read
Unknown profilePreserve evidence; do not prescribe stack repair
Mutation requiredPropose separately and wait for authorization

Comments

Markdown supported