MCP Server for Kubernetes Support Bundles

Name: MCP Server for Kubernetes Support Bundles
Author: chris-sanders

@chris-sanders

Repository

About MCP Server for Kubernetes Support Bundles

No overview available yet

Basic information

Config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "troubleshoot-mcp-server": {
      "command": "uv",
      "args": [
        "venv",
        "-p",
        "python3.13",
        ".venv"
      ]
    }
  }
}

Tools

No tools detected

We auto-extract tools from the README. The maintainer can list them under a ## Tools heading to populate this section.

Overview

What is MCP Server for Kubernetes Support Bundles?

MCP Server for Kubernetes Support Bundles is a Model Context Protocol (MCP) server that lets AI models analyze and troubleshoot Kubernetes clusters by exploring support bundles generated by the Troubleshoot tool. It provides tools for bundle management, kubectl command execution, and file operations within the bundle.

How to use MCP Server for Kubernetes Support Bundles?

Run the server with Podman or install manually with Python 3.13 and UV. Set the authentication token via the SBCTL_TOKEN environment variable. For stateless deployments, enable Single Bundle Mode with MCP_SINGLE_BUNDLE_MODE=true and PRESERVE_BUNDLES=true. Invoke tools like initialize_bundle, kubectl, list_files, read_file, and grep_files over the MCP protocol.

Key features of MCP Server for Kubernetes Support Bundles

Bundle management: initialize and manage support bundles
Execute kubectl commands against the bundle’s API server
Navigate, search, and read files within the bundle
Token‑based authentication for secure access
Containerized deployment with Podman (development and production variants)
Single Bundle Mode for ephemeral/serverless environments

Use cases of MCP Server for Kubernetes Support Bundles

Diagnose Kubernetes cluster issues by analyzing support bundles
Run kubectl commands on a static bundle without a live cluster
Browse and search cluster resource definitions and logs inside a bundle
Integrate into automated workflows (Temporal, serverless, container‑per‑request)
Enable AI assistants to troubleshoot Kubernetes from bundle data

FAQ from MCP Server for Kubernetes Support Bundles

How do I enable Single Bundle Mode?

Set MCP_SINGLE_BUNDLE_MODE=true and PRESERVE_BUNDLES=true, and optionally configure MCP_BUNDLE_STORAGE for a persistent directory. This mode treats the presence of a bundle on disk as the source of truth, allowing automatic restoration after restarts.

What are the runtime requirements?

Python 3.13, the kubectl command‑line tool, and the sbctl tool for bundle management. A token (set as SBCTL_TOKEN or REPLICATED environment variable) is required for authentication. The Podman container includes all dependencies.

How do I run the server with Podman?

Build the image with ./scripts/build.sh, then run with podman run -i --rm -v "/path/to/bundles:/data/bundles" -e SBCTL_TOKEN="your-token" troubleshoot-mcp-server-dev:latest.

What tools does the server expose to AI models?

initialize_bundle, kubectl, list_files, read_file, and grep_files. These allow bundle initialization, command execution, and file exploration within the bundle.

Can I use multiple bundles at the same time?

In default mode, multiple bundles can be managed. Single Bundle Mode enforces that only one bundle exists at a time, preventing state confusion in ephemeral deployments.