Overview
Containerized MCP server deployment platform for simplified running and management.
Setup
Run with npx:
curl -fsSL https://toolhive.io/install.sh | bashConfiguration
ToolHive CLI with environment variables for API keysDocumentation
ToolHive
Overview
ToolHive is a containerized MCP server deployment platform that simplifies running, managing, and scaling MCP servers. It provides a unified interface for deploying MCP servers as containers, handling dependency management, networking, and lifecycle management automatically.
ToolHive addresses one of the biggest challenges with MCP servers — deployment complexity. Instead of manually installing and configuring each MCP server, ToolHive provides a catalog of pre-built containers that can be deployed with a single command. It also provides a management layer for monitoring, updating, and scaling MCP server deployments.
Features
- Containerized Deployment: Run MCP servers in isolated containers
- One-Command Deploy: Deploy any MCP server from the catalog with a single command
- Dependency Management: Automatic handling of all dependencies
- Network Isolation: Secure sandboxed environment for each server
- Centralized Management: Manage all MCP servers from one interface
- Auto-Updates: Automatic updates for security patches and new features
- Resource Limits: Configure CPU, memory, and network limits per server
- Logging & Monitoring: Built-in logging and health checks
- Multi-Platform: Works on macOS, Linux, and Windows
Installation
Using the ToolHive CLI
# Install ToolHive
curl -fsSL https://toolhive.io/install.sh | bash
# Or using Homebrew (macOS)
brew install toolhive
Using Docker Directly
docker pull ghcr.io/toolhive/toolhive:latest
Quick Start
Deploy an MCP Server
# List available MCP servers
toolhive list
# Deploy GitHub MCP
toolhive run github
# Deploy with custom configuration
toolhive run github \
--name my-github \
--env GITHUB_TOKEN=ghp_xxx \
--memory 512m \
--cpu 1.0
Connect to Claude Desktop
ToolHive automatically configures Claude Desktop:
# Configure Claude Desktop to use ToolHive servers
toolhive configure claude-desktop
# Or manually add to config:
# ~/.config/Claude/claude_desktop_config.json
{
"mcpServers": {
"github": {
"command": "toolhive",
"args": ["run", "github", "--stdio"]
}
}
}
Available MCP Servers
ToolHive maintains a catalog of pre-built MCP server containers:
| Server | Description |
|---|---|
github | GitHub repository management |
postgresql | PostgreSQL database access |
notion | Notion workspace integration |
slack | Slack messaging and channels |
filesystem | Local file system access |
brave-search | Web search via Brave API |
fetch | Web page fetching and conversion |
memory | Knowledge graph memory |
sequential-thinking | Structured problem solving |
time | Time and timezone utilities |
Configuration
Environment Variables
# Set API keys and credentials
toolhive env set GITHUB_TOKEN ghp_xxx
toolhive env set NOTION_TOKEN secret_xxx
toolhive env set SLACK_BOT_TOKEN xoxb-xxx
Resource Limits
# Deploy with resource constraints
toolhive run postgresql \
--memory 1g \
--cpu 2.0 \
--disk 10g
Network Configuration
# Expose MCP server on a specific port
toolhive run github --port 3000
# Allow specific network access
toolhive run github --network allow-https
Advanced Features
Custom MCP Servers
# Build a custom MCP server container
toolhive build ./my-mcp-server -t my-registry/my-mcp:latest
# Run your custom server
toolhive run my-registry/my-mcp:latest
Server Groups
# Create a group of related servers
toolhive group create dev-tools github postgresql filesystem
# Deploy all servers in a group
toolhive group deploy dev-tools
Health Monitoring
# Check server health
toolhive health check
# View server logs
toolhive logs github --follow
# Restart a server
toolhive restart github
Auto-Scaling
# Enable auto-scaling for high-traffic servers
toolhive scale github --replicas 3
Examples
Full Dev Environment Setup
# Deploy a complete development toolchain
toolhive group create dev-stack \
github \
postgresql \
filesystem \
brave-search \
fetch
# Configure all for Claude Desktop
toolhive group configure dev-stack claude-desktop
# Start all servers
toolhive group start dev-stack
Production Deployment
# Deploy with production settings
toolhive run github \
--name prod-github \
--env GITHUB_TOKEN=$GITHUB_TOKEN \
--memory 2g \
--cpu 2.0 \
--replicas 3 \
--health-check-interval 30s \
--auto-update
# Set up monitoring
toolhive monitoring enable --endpoint https://monitoring.example.com
Pros
- ✅ Simplifies MCP server deployment significantly
- ✅ Containerized for security and isolation
- ✅ Large catalog of pre-built servers
- ✅ Automatic dependency resolution
- ✅ Centralized management interface
- ✅ Built-in monitoring and logging
- ✅ Auto-update support
- ✅ Multi-platform support
Cons
- ❌ Adds a layer of complexity (Docker required)
- ❌ Resource overhead from containerization
- ❌ Some servers may not be in the catalog yet
- ❌ Learning curve for ToolHive-specific commands
- ❌ Requires Docker installed and running
When to Use
- Deploying multiple MCP servers — Centralized management
- Production MCP deployments — Container isolation and scaling
- Team environments — Standardized server deployments
- Security-conscious deployments — Sandboxed execution
- Rapid prototyping — One-command server deployment
