About Claude Mcp Server Gateway
A Python MCP gateway for Claude Desktop, Claude Code, Cline, and Roo that loads tools on demand instead of loading every server’s tool definitions at startup. It provides one entry point for multiple configured MCP servers.
Key features
- 01Load MCP tools on demand
- 02List configured servers
- 03Forward tool calls
- 04Start backend servers temporarily
- 05Support multiple client configs
Use cases
- 01Reduce tool definitions loaded at startup
- 02Call a specific MCP server for a task
- 03Manage multiple servers through one entry point
MCP support
The gateway acts as one MCP server, exposing server listing, tool loading, and tool execution to clients. Backend MCP servers start only when needed, and connections are temporary.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEClaude MCP server gateway
A Python MCP server gateway for Claude Desktop, Claude Code, Cline and Roo. Instead of loading all MCP tools at startup, it loads them on demand.
Works with: Claude Desktop (app), Claude Code (terminal), Cline and Roo (VSCode extensions)
Compatibility
| Platform | Config File | Supported |
|---|---|---|
| Claude Desktop | claude_desktop_config.json | Yes |
| Claude Code | .claude.json or .mcp.json | Yes |
| Cline (VSCode) | cline_mcp_settings.json | Yes |
The MCP server problem
When you connect multiple MCP servers to Claude:
- GitHub MCP server: 51 tools
- Slack MCP server: 20 tools
- Google Drive MCP: 15 tools
- Filesystem MCP: 12 tools
That's 100+ MCP tool definitions loaded before you even start. Each tool's JSON schema eats up tokens.
What this Claude MCP server gateway does
The gateway acts as a single MCP server with 3 tools:
list_available_mcps- shows configured MCP serversload_mcp_tools- loads tools from a specific MCP servercall_mcp_tool- executes the tool
Model Context Protocol servers only start when Claude actually needs them.
Setup
Requirements
- Python 3.11+ (for MCP Python SDK)
- Claude Desktop, Claude Code, or Cline/Roo
- MCP servers you want to use
Install
git clone https://github.com/bzsasson/claude-mcp-server-gateway.git
cd claude-mcp-server-gateway
python3.11 -m venv .venv
source .venv/bin/activate # Windows: .venv\Scripts\activate
pip install mcp python-dotenv
Configure API keys
cp .env.example .env
# Edit .env with your tokens
Claude Desktop MCP server configuration
Add to your Claude Desktop config:
Mac: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
Linux: ~/.config/Claude/claude_desktop_config.json
{
"mcpServers": {
"gateway": {
"command": "/path/to/.venv/bin/python",
"args": ["/path/to/claude-mcp-server-gateway/dcl_wrapper.py"]
}
}
}
Restart Claude Desktop.
How to add MCP server to Claude Code
Claude Code MCP server setup uses a similar config with one additional field.
User-level configuration (all projects)
Edit ~/.claude.json:
{
"mcpServers": {
"gateway": {
"type": "stdio",
"command": "/path/to/.venv/bin/python",
"args": ["/path/to/claude-mcp-server-gateway/dcl_wrapper.py"]
}
}
}
Project-level configuration
For project-specific MCP servers, create .mcp.json in your project root:
{
"mcpServers": {
"gateway": {
"type": "stdio",
"command": "/path/to/.venv/bin/python",
"args": ["/path/to/claude-mcp-server-gateway/dcl_wrapper.py"]
}
}
}
Claude Code will prompt for approval when using project-scoped servers.
Alternative: Claude Code CLI
# Add the gateway using CLI
claude mcp add gateway --scope user \
--command /path/to/.venv/bin/python \
-- /path/to/claude-mcp-server-gateway/dcl_wrapper.py
# Verify it's added
claude mcp list
Cline/Roo MCP server configuration
For Cline and Roo (VSCode extensions), the MCP server gateway configuration goes in:
Mac: ~/Library/Application Support/Code/User/globalStorage/saoudrizwan.claude-dev/settings/cline_mcp_settings.json
Windows: %APPDATA%/Code/User/globalStorage/saoudrizwan.claude-dev/settings/cline_mcp_settings.json
Linux: ~/.config/Code/User/globalStorage/saoudrizwan.claude-dev/settings/cline_mcp_settings.json
{
"mcpServers": {
"gateway": {
"command": "/path/to/.venv/bin/python",
"args": ["/path/to/claude-mcp-server-gateway/dcl_wrapper.py"],
"disabled": false
}
}
}
Restart VSCode after updating.
How it works
- Claude starts - sees gateway (3 tools)
- You ask about GitHub - Claude calls
list_available_mcps - Claude sees GitHub available - calls
load_mcp_tools("github") - Gateway returns GitHub's tool list
- Claude picks the right tool - calls
call_mcp_tool("github", "search_repositories", {...}) - Gateway spins up GitHub MCP, runs the tool, returns results
Adding MCP servers to the gateway
Edit dcl_wrapper.py and add to the MCP_SERVERS dict:
MCP_SERVERS = {
"your-server": {
"command": "npx",
"args": ["-y", "your-mcp-package"],
"env": {
"API_KEY": os.getenv("YOUR_API_KEY", "")
},
"description": "What it does"
}
}
Currently configured servers
- github
- slack
- google-drive
- filesystem
- postgres
- memory
- brave-search
(Check dcl_wrapper.py for the full list)
Limitations
- Small delay on first tool use (server startup)
- Each MCP server connection is temporary
- No persistence between calls
Troubleshooting
Gateway not appearing
- Check Python path in config points to virtual environment
- Verify
dcl_wrapper.pypath is correct - Restart Claude Desktop/Code/VSCode completely
MCP server connection errors
- Check API keys in
.envfile - Verify MCP server package is installed
- Check gateway logs for errors
Claude Code specific issues
- Make sure to include
"type": "stdio"in config - For project configs, approve when prompted
- Use
claude mcp listto verify installation
Files
dcl_wrapper.py- The gateway server.env- Your API keysrequirements.txt- Python dependencies
License
MIT
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.