apiiro-mongo-mcp by arielolin - MCP Server

MongoDB kubectl MCP Server

An MCP (Model Context Protocol) server that connects to MongoDB running in Kubernetes via kubectl exec. Handles gcloud authentication, permission escalation, and client/namespace selection.

Features

🔐 gcloud Authentication: Automatic gcloud auth setup
⬆️ Permission Escalation: Escalate to production permissions
🎯 Multi-Context Support: Works with multiple Kubernetes contexts
👥 Client Namespaces: Select client by namespace
🍃 MongoDB Operations: Query MongoDB via kubectl exec

Prerequisites

Node.js 18+
kubectl configured with access to your clusters
gcloud CLI installed and configured
MongoDB deployment named lim-mongo in your namespaces

Installation

Follow these steps to install all prerequisites and set up the MCP server:

1. Install Node.js

macOS (using Homebrew):

brew install node@18
# or for latest LTS
brew install node

macOS (using nvm - recommended):

# Install nvm if not already installed
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.0/install.sh | bash
source ~/.bashrc  # or restart terminal

# Install Node.js 18+
nvm install 18
nvm use 18

Linux:

# Using NodeSource repository
curl -fsSL https://deb.nodesource.com/setup_18.x | sudo -E bash -
sudo apt-get install -y nodejs

Verify installation:

node --version  # Should be 18 or higher
npm --version

2. Install kubectl

macOS (using Homebrew):

brew install kubectl

macOS (using curl):

curl -LO "https://dl.k8s.io/release/$(curl -L -s https://dl.k8s.io/release/stable.txt)/bin/darwin/arm64/kubectl"
chmod +x kubectl
sudo mv kubectl /usr/local/bin/

Linux:

curl -LO "https://dl.k8s.io/release/$(curl -L -s https://dl.k8s.io/release/stable.txt)/bin/linux/amd64/kubectl"
chmod +x kubectl
sudo mv kubectl /usr/local/bin/

Verify installation:

kubectl version --client

3. Install gcloud CLI

macOS (using Homebrew - recommended):

brew install --cask google-cloud-sdk

macOS (using installer script):

# Download and install
curl https://sdk.cloud.google.com | bash

# Restart your shell or run:
exec -l $SHELL

# Initialize gcloud
gcloud init

Linux:

# Add Cloud SDK distribution URI as a package source
echo "deb [signed-by=/usr/share/keyrings/cloud.google.gpg] https://packages.cloud.google.com/apt cloud-sdk main" | sudo tee -a /etc/apt/sources.list.d/google-cloud-sdk.list

# Import the Google Cloud Platform public key
curl https://packages.cloud.google.com/apt/doc/apt-key.gpg | sudo apt-key --keyring /usr/share/keyrings/cloud.google.gpg add -

# Update and install
sudo apt-get update && sudo apt-get install google-cloud-cli

Verify installation:

gcloud --version

4. Configure gcloud Authentication

# Login to your Google account
gcloud auth login

# Set your default project
gcloud config set project apiiro

# Verify authentication
gcloud auth list

5. Configure kubectl for GKE Clusters

Get cluster credentials from GKE:

Option A: Get credentials for a specific cluster

# Authenticate with the cluster
gcloud container clusters get-credentials CLUSTER_NAME \
    --region REGION \
    --project PROJECT_ID

# Example:
gcloud container clusters get-credentials my-cluster \
    --region us-central1 \
    --project my-project-id

Option B: List and connect to available clusters

# List all clusters in your project
gcloud container clusters list

# Get credentials for a cluster (will add it to your kubeconfig)
gcloud container clusters get-credentials CLUSTER_NAME --region REGION

Verify kubectl access:

# List contexts (you should see your GKE cluster)
kubectl config get-contexts

# Test connection
kubectl get nodes

For multiple clusters:

# Get credentials for each cluster
gcloud container clusters get-credentials cluster-1 --region us-central1
gcloud container clusters get-credentials cluster-2 --region us-east1

# List all contexts
kubectl config get-contexts

Add R&D Context (Required for Apiiro):

# Set the R&D project
gcloud config set project apiiro-rnd

# Get credentials for the R&D self-service cluster
gcloud container clusters get-credentials apiiro-rnd-self-service-us \
    --region us-central1 \
    --project apiiro-rnd

# Verify the R&D context was added
kubectl config get-contexts | grep apiiro-rnd

# Expected context name: gke_apiiro-rnd_us-central1_apiiro-rnd-self-service-us

6. Install Project Dependencies

Clone or navigate to the project directory, then:

# Install Node.js dependencies
npm install

# Build the project
npm run build

7. Verify Complete Setup

Run these commands to verify everything is configured correctly:

# Check Node.js
node --version

# Check gcloud authentication
gcloud auth list

# Check kubectl can access clusters
kubectl config get-contexts

# Verify R&D context is available
kubectl config get-contexts | grep apiiro-rnd

# Test cluster access (switch to R&D context first)
kubectl config use-context gke_apiiro-rnd_us-central1_apiiro-rnd-self-service-us
kubectl get namespaces

# Verify project build
ls -la dist/index.js

You're now ready to use the MCP server! Proceed to the Usage section to get started.

Usage

1. Setup (Run First!)

Use the setup tool to authenticate and select your client:

{
  "escalate": true,
  "context": "gke-project-region-cluster",
  "client": "client-name-namespace"
}

Or run interactively - the tool will list available options if you don't specify them.

2. Query MongoDB

Once setup is complete, use MongoDB tools:

mongo_query: Execute MongoDB queries

{
  "query": "db.users.find().limit(5)"
}

mongo_eval: Evaluate JavaScript expressions
```
{
  "expression": "db.stats()"
}
```

3. List Options

list_contexts: List available Kubernetes contexts
list_namespaces: List namespaces in a context
get_status: Check current configuration

MCP Configuration

Claude Desktop

The server has been configured in your Claude Desktop config at: ~/Library/Application Support/Claude/claude_desktop_config.json

If you need to add it manually:

{
  "mcpServers": {
    "mongo-kubectl": {
      "command": "node",
      "args": ["/Users/arielolin/mongo-kubectl-mcp/dist/index.js"]
    }
  }
}

After adding, restart Claude Desktop for the changes to take effect.

Cursor

Add to your MCP settings (e.g., ~/.cursor/mcp.json or similar):

{
  "mcpServers": {
    "mongo-kubectl": {
      "command": "node",
      "args": ["/path/to/mongo-kubectl-mcp/dist/index.js"]
    }
  }
}

Architecture

Setup Tool: Handles gcloud auth → escalation → context selection → namespace selection
State Management: Saves configuration to config.json for persistence
kubectl Exec: Uses kubectl exec -it deploy/lim-mongo with selected context/namespace
mongosh: Executes MongoDB commands via mongosh shell

Troubleshooting

No contexts found: Make sure kubectl is configured and contexts are available
Authentication errors: Run gcloud auth login manually if needed
Namespace not found: Verify the namespace exists in the selected context
MongoDB connection fails: Check that lim-mongo deployment exists in the namespace