☕ Looking for Java? See Java Prerequisites | 📘 Looking for TypeScript? See TypeScript Prerequisites

Prerequisites¶

What you need before building MCP Mesh agents with Python

MCP Mesh supports Python, Java, and TypeScript. Choose the language that fits your needs—or use multiple languages in the same mesh.

Windows Users¶

meshctl and mcp-mesh-registry require a Unix-like environment on Windows:

WSL2 (recommended) - Full Linux environment
Git Bash - Lightweight option

Alternatively, use Docker Desktop for containerized development.

Local Development¶

For developing and testing agents locally.

Python 3.11+¶

# Check version
python3 --version  # Need 3.11+

# Install if needed
brew install python@3.11          # macOS
sudo apt install python3.11       # Ubuntu/Debian

Virtual Environment (Recommended)¶

Create a virtual environment at your project root (where you run meshctl). All agents share this single venv - do not create separate venvs inside agent folders.

Note: meshctl is a Go binary that auto-detects .venv in the current directory. You only need to activate the venv for pip commands - meshctl commands work without activation.

# At project root - create venv (one-time setup)
python3.11 -m venv .venv

# Activate only when using pip
source .venv/bin/activate         # macOS/Linux
.venv\Scripts\activate            # Windows
pip install --upgrade pip

MCP Mesh SDK¶

pip install "mcp-mesh>=0.8,<0.9"

# Verify
python -c "import mesh; print('Ready!')"

Quick Start (Python)¶

# 1. Create venv and install SDK (one-time setup)
python3.11 -m venv .venv
source .venv/bin/activate    # Only needed for pip
pip install --upgrade pip
pip install "mcp-mesh>=0.8,<0.9"
deactivate                   # Can deactivate after pip install

# 2. Scaffold agents - meshctl auto-detects .venv (no activation needed)
meshctl scaffold --name hello --agent-type basic
meshctl scaffold --name assistant --agent-type llm-agent

# 3. Run agent - meshctl uses .venv/bin/python automatically
meshctl start hello/main.py --debug

Docker Deployment¶

For containerized deployments.

Docker & Docker Compose¶

# Check installation
docker --version
docker compose version

MCP Mesh Images¶

Image	Description
`mcpmesh/registry:0.9`	Registry service
`mcpmesh/python-runtime:0.9`	Python runtime with SDK
`mcpmesh/java-runtime:0.9`	Java runtime with SDK
`mcpmesh/typescript-runtime:0.9`	TypeScript runtime with SDK

# Pull images
docker pull mcpmesh/registry:0.9
docker pull mcpmesh/python-runtime:0.9
docker pull mcpmesh/java-runtime:0.9
docker pull mcpmesh/typescript-runtime:0.9

Generate Docker Compose¶

meshctl scaffold --compose              # Basic stack
meshctl scaffold --compose --observability  # With Grafana/Tempo

Kubernetes Deployment¶

For production Kubernetes clusters.

kubectl¶

kubectl version --client

Helm 3+¶

helm version

MCP Mesh Helm Charts¶

Available from OCI registry (no helm repo add needed):

Chart	Description
`oci://ghcr.io/dhyansraj/mcp-mesh/mcp-mesh-core`	Registry + DB + Observability
`oci://ghcr.io/dhyansraj/mcp-mesh/mcp-mesh-agent`	Deploy agents

# Install core infrastructure
helm install mcp-core oci://ghcr.io/dhyansraj/mcp-mesh/mcp-mesh-core \
  --version 0.9.0 \
  -n mcp-mesh --create-namespace

# Deploy an agent
helm install my-agent oci://ghcr.io/dhyansraj/mcp-mesh/mcp-mesh-agent \
  --version 0.9.0 \
  -n mcp-mesh \
  -f helm-values.yaml

Cluster Options¶

Minikube - Local testing
Kind - Lightweight local clusters
Cloud - GKE, EKS, AKS for production

Version Compatibility¶

Component	Minimum	Recommended
Python	3.11	3.12
Docker	20.10	Latest
Kubernetes	1.25	1.28+
Helm	3.10	3.14+