akkireddy-challa/k8s-mcp-server

> Production-ready Kubernetes diagnostics and cluster observability MCP server for AI agents.
---
What is this?
k8s-mcp-server is a Model Context Protocol (MCP) server providing automated, read-only diagnostic tools for inspecting Kubernetes clusters. It allows AI agents to troubleshoot pod crashes, inspect ingress routes, and analyze cluster-wide warning events safely.
---
Available Tools
| Tool | Category | Description |
|---|---|---|
| get_cluster_nodes | Infrastructure | Lists nodes, readiness states, roles, and kubelet versions. |
| get_pod_diagnostics | Workloads | Detects abnormal pod phases, restart loops, and CrashLoopBackOffs. |
| get_pod_logs | Observability | Retrieves container stdout/stderr logs with tail limits and crash inspection. |
| list_warning_events | Diagnostics | Aggregates FailedScheduling, FailedMount, and BackOff warning events. |
| list_ingresses | Networking | Audits HTTP routing rules, host headers, and TLS certificates. |
---
Quick Start
Prerequisites
- Python 3.11+
- Active Kubernetes cluster access (~/.kube/config or in-cluster ServiceAccount)
- Read-only RBAC privileges (get, list on core and networking API groups)
Run Locally
Claude Desktop Configuration
Add to claude_desktop_config.json:
---
Security Model
- **Strictly Read-Only**: Enforces GET and LIST operations only.
- **Zero-Secret Exposure**: Secrets and config maps are never inspected or surfaced to LLM context.
- **Flexible Auth**: Supports standard ~/.kube/config context or in-cluster pod ServiceAccounts.
---
License
MIT License. See LICENSE for details.
---
Built by Akkireddy Challa — Platform Engineer at Telia, Stockholm.