Skip to content

Quick start

Three things have to be in place: Ollama running, a config file with at least one MCP server, and the bridge itself.

1. Start Ollama

ollama serve

The bridge checks Ollama's health before binding its own port, so if this step is missing you get a clear error instead of a server that half-starts.

2. Write an mcp-config.json

The repository ships a working example that launches a small mock weather server, so you can exercise the bridge without any external MCP server:

mcp-config.json
{
  "mcpServers": {
    "weather": {
      "command": "uv",
      "args": ["--directory", "./mock-weather-mcp-server", "run", "main.py"],
      "env": {
        "MCP_LOG_LEVEL": "ERROR"
      }
    }
  }
}

Paths are relative to the config file

Not to the directory you run the bridge from. ./mock-weather-mcp-server resolves next to mcp-config.json, which means you can move the config around without rewriting every path. The same rule drives ${workspaceFolder}.

3. Start the bridge

ollama-mcp-bridge

Defaults: config ./mcp-config.json, host 0.0.0.0, port 8000, Ollama at http://localhost:11434. On startup it connects every configured MCP server, collects their tools and logs what it found.

4. Send a request

Point any Ollama client at the bridge instead of Ollama.

from ollama import Client

client = Client(host="http://localhost:8000")

response = client.chat(
    model="qwen3:0.6b",
    messages=[{"role": "user", "content": "What is the weather like in Paris today?"}],
)
print(response["message"]["content"])

curl -N -X POST http://localhost:8000/api/chat \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3:0.6b",
    "messages": [
      {"role": "system", "content": "You are a weather assistant."},
      {"role": "user", "content": "What is the weather like in Paris today?"}
    ],
    "think": true,
    "stream": true,
    "options": {"temperature": 0.7, "top_p": 0.9}
  }'

The model will ask for the weather tool, the bridge will execute it against the MCP server, feed the result back, and return the finished answer — all inside that one request.

  • Understand the loop

    What the bridge does between receiving your request and answering it.

    How it works

  • Connect real servers

    stdio, StreamableHTTP and SSE, plus per-server tool filtering.

    MCP servers