# Transport Options for GitLab MCP Server
GitLab MCP Server supports three transport types for communication with MCP clients:
## 1. stdio Transport
The stdio transport is the simplest option and is ideal for direct integration with Claude Desktop and other MCP clients that support stdio communication.
```bash
# Run with stdio transport (default)
gitlab-mcp stdio
# Or using Python directly
python -m gitlab_clone_mcp_server stdio
```
## 2. Server-Sent Events (SSE) Transport
The SSE transport provides a web-based interface using Server-Sent Events, which is suitable for web applications and services.
```bash
# Run with SSE transport
gitlab-mcp sse --host localhost --port 8000
# Or using Python directly
python -m gitlab_clone_mcp_server sse --host localhost --port 8000
```
## 3. Streamable HTTP Transport
The Streamable HTTP transport is the newest and recommended transport for production deployments, offering better scalability and reliability.
```bash
# Run with Streamable HTTP transport
gitlab-mcp streamable-http --host localhost --port 8000
# Or using Python directly
python -m gitlab_clone_mcp_server streamable-http --host localhost --port 8000
```
## Mounting to Existing ASGI Applications
You can mount the GitLab MCP Server to an existing ASGI application:
```python
from starlette.applications import Starlette
from starlette.routing import Mount
import contextlib
from gitlab_clone_mcp_server.server import mcp
# Create a Starlette app
app = Starlette()
# Create a combined lifespan
@contextlib.asynccontextmanager
async def lifespan(app):
async with contextlib.AsyncExitStack() as stack:
await stack.enter_async_context(mcp.session_manager.run())
yield
# Mount the MCP server
app.router.routes.append(Mount("/mcp", mcp.streamable_http_app()))
app.router.routes.append(Mount("/sse", mcp.sse_app()))
# Set the lifespan
app.lifespan = lifespan
```
## Claude Desktop Configuration
To use GitLab MCP Server with Claude Desktop, add the following to your Claude Desktop config:
```json
{
"mcpServers": {
"gitlab": {
"command": "gitlab-mcp",
"args": ["stdio"],
"env": {
"GITLAB_URL": "https://gitlab.com",
"GITLAB_TOKEN": "your_gitlab_token_here"
}
}
}
}
```
For HTTP-based transports, use:
```json
{
"mcpServers": {
"gitlab": {
"url": "http://localhost:8000/mcp",
"headers": {
"GITLAB_TOKEN": "your_gitlab_token_here"
}
}
}
}
```
## MCPToolset Integration
For applications using MCPToolset:
```python
import os
from mcp_toolset import MCPToolset, SseServerParams
gitlab_tools = MCPToolset(
connection_params=SseServerParams(
url="http://localhost:8000/sse",
headers={"GITLAB_TOKEN": os.getenv("GITLAB_TOKEN")},
)
)
async def gitlab_operations():
async with gitlab_tools:
projects = await gitlab_tools.call_tool("list_projects", {"per_page": 20})
return projects
```
### Available Connection Parameters
- **SseServerParams**: For Server-Sent Events transport
- **StreamableHttpServerParams**: For Streamable HTTP transport
- **StdioServerParams**: For stdio transport (local execution)
```python
# SSE Transport
from mcp_toolset import SseServerParams
sse_params = SseServerParams(
url="http://localhost:8000/sse",
headers={"GITLAB_TOKEN": os.getenv("GITLAB_TOKEN")}
)
# Streamable HTTP Transport
from mcp_toolset import StreamableHttpServerParams
http_params = StreamableHttpServerParams(
url="http://localhost:8000/mcp",
headers={"GITLAB_TOKEN": os.getenv("GITLAB_TOKEN")}
)
# Stdio Transport (for local execution)
from mcp_toolset import StdioServerParams
stdio_params = StdioServerParams(
command="gitlab-mcp",
args=["stdio"],
env={"GITLAB_TOKEN": os.getenv("GITLAB_TOKEN")}
)
```
