gke-workload-troubleshooting

Find why your GKE workload fails and get a proposed fix.

Kubernetes troubleshooting

What it does

Diagnoses GKE application workload failures and proposes fixes without mutating the live cluster.

When to use it

Use when pods fail to start or repeatedly crash.

Give it a project containing a failing GKE workload; it diagnoses the cause and opens or updates a GitHub pull request with a fix.

What you provide

This skill

GitHub

Write

Opens or updates GitHub PRs (irreversible)

GKE cluster

Read

Reads GKE workload diagnostics

Cloud Logging

Read

Queries Cloud Logging diagnostics

active SETTINGS.md

Read

Reads active SETTINGS.md

Requires access to GitHub to check for an existing branch or pull request and create or update one.

kubectl enables live inspection of workload state, events, logs, endpoints, and network policies; dry-run fallback is available when commands fail.

gcloud enables cluster credential retrieval and Cloud Logging queries; offline analysis is used if live execution fails.

Live access to a GKE cluster enables direct diagnostics, but an unreachable or nonexistent cluster triggers dry-run analysis.