Talk to Your Kubernetes Cluster Like It's a Teammate

Stop context-switching between kubectl, Grafana dashboards, and log aggregators every time something breaks in production. The Metoro MCP Server bridges your Kubernetes observability data directly into Claude Desktop, so you can debug issues and explore your cluster using natural language.

The Problem Every K8s Developer Knows

You're deep in code when Slack lights up—something's wrong in production. Now you're juggling:

• kubectl commands to check pod status
• Grafana to see if metrics look normal
• Log aggregation tools to hunt for errors
• Mental mapping between services and their dependencies

By the time you piece together what's happening, you've lost 30 minutes and your flow state.

Ask Your Cluster Directly

Instead of this tool-hopping dance, just ask Claude:

• "Which pods are consuming the most memory right now?"
• "Show me error rates for the payment service over the last hour"
• "Are there any failing health checks in the auth namespace?"
• "What's the network traffic pattern between my frontend and backend services?"

The MCP server connects Claude to your Metoro observability data, giving you conversational access to the same eBPF-based telemetry that powers your dashboards—but without leaving your chat interface.

Real Debugging Workflows

Incident Response: When alerts fire, ask Claude to correlate metrics across services instead of manually checking each dashboard. "What changed in the last 10 minutes that could be causing these 500 errors?"

Performance Investigation: Skip the dashboard rabbit holes. "Which database queries are taking longer than usual?" gets you straight to the bottleneck.

Capacity Planning: Instead of exporting CSV files and building spreadsheets, ask "What's our resource utilization trend for the recommendations service?"

Deployment Verification: After pushing changes, verify everything's healthy with "How do the error rates and response times look for the services I just deployed?"

Setup Takes 5 Minutes

Since you probably have Go installed already:

git clone https://github.com/metoro-io/metoro-mcp-server.git
cd metoro-mcp-server
go build -o metoro-mcp-server

Add the server to your Claude Desktop config with your Metoro auth token (or use their demo cluster to try it out first). Restart Claude, and you're ready to start asking questions about your infrastructure.

The server connects to Metoro's APIs that process eBPF telemetry from your cluster—so you get the same deep observability data you'd see in their dashboard, just accessible through conversation.

Why This Approach Works

Traditional observability tools make you think like a dashboard—you need to know which metrics to check and how to correlate them. This flips that model: describe what you're investigating in plain English, and let the AI figure out which data points matter.

You're not replacing your existing tools, just adding a more intuitive interface that understands context. When Claude suggests checking specific metrics or logs, you can still jump into your familiar tools with focused intent.

The Metoro MCP Server runs locally and connects to your existing Metoro deployment, so there's no additional infrastructure to manage or new observability agents to install.

Try the demo cluster first to see how conversational K8s debugging feels, then connect it to your actual infrastructure. Your future self will thank you the next time production decides to have opinions at 2 AM.