# QuantConnect/mcp-server [Health: Active]

**Category:** 💰 Finance & Fintech  
**Repository:** https://github.com/QuantConnect/mcp-server  
**GitHub Stars:** 77  
**Views:** 3  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/quantconnect-mcp-server

## Description
A Dockerized Python MCP server that bridges your local AI (e.g., Claude Desktop, etc) with the QuantConnect API—empowering you to create projects, backtest strategies, manage collaborators, and deploy live-trading workflows directly via natural-language prompts.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "mcp-server": {
    "command": "npx",
    "args": ["-y","@modelcontextprotocol/inspector"],
    "env": {
      "QUANTCONNECT_USER_ID": "",
      "QUANTCONNECT_API_TOKEN": "",
      "AGENT_NAME": ""
    }
  }
}
```

**Requires environment variables:** `QUANTCONNECT_USER_ID`, `QUANTCONNECT_API_TOKEN`, `AGENT_NAME` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What QuantConnect/mcp-server MCP server does

QuantConnect/mcp-server MCP server gives an MCP-compatible AI client access to QuantConnect account and cloud-platform operations. It can create, inspect, modify, and delete projects; manage project collaborators; read and edit project files; and compile algorithm code. The available tools also cover backtests, optimization jobs, live algorithms, Object Store files, LEAN version information, and several code-assistance operations.

The server is intended for users who want an AI assistant to work with QuantConnect projects through natural-language requests. Depending on the operation, an assistant can update strategy code, check syntax, run initialization checks, start a backtest, inspect results, or manage a live-trading deployment. Destructive operations are included, such as deleting projects, files, backtests, optimizations, and Object Store files, so client permissions and prompts should be handled carefully.

## How it works

The distribution runs as a Docker image and communicates with a local MCP client over the client’s configured process connection. Claude Desktop is the documented example. The container authenticates to QuantConnect using `QUANTCONNECT_USER_ID` and `QUANTCONNECT_API_TOKEN`. `AGENT_NAME` identifies the request source and is useful when multiple agents operate at the same time.

QuantConnect/mcp-server MCP server exposes 64 listed tools. Project tools provide access to projects, collaborators, nodes, files, compilation, and code checks. Research and execution tools cover backtests, charts, orders, insights, optimization, and live algorithm status. Other tools search QuantConnect content, access Object Store data, authorize brokerage or data-provider connections, and report the running or latest MCP server version.

## Setup and configuration

Docker Desktop and an MCP client are required for the documented local setup. In Claude Desktop, add a server entry that runs the `quantconnect/mcp-server` Docker image with interactive input and removes the container after exit. Pass the two QuantConnect credentials and an agent name through the container environment.

The image supports both `linux/amd64` and `linux/arm64`. ARM-based systems, including Apple M-series machines, should select the ARM platform when configuring Docker. The README also documents pulling the image with `docker pull quantconnect/mcp-server` to update the local copy.

After changing the Claude Desktop configuration, restart the client. QuantConnect’s documentation links to its API-token request instructions for obtaining credentials.

## Limitations and notes

This repository is marked deprecated. QuantConnect states that the preferred MCP experience is the one embedded in VS Code, so new deployments should evaluate that option before using this Docker-based repository.

The material documents Claude Desktop as a local client example, but does not provide a complete compatibility list for every MCP client. It also does not specify pricing, repository licensing, or the access requirements for individual QuantConnect account features. Live-trading tools can authorize connections, create deployments, send commands, stop algorithms, and liquidate them; use those capabilities only with appropriate safeguards.

QuantConnect/mcp-server MCP server is most relevant when the AI client needs to perform operations against an existing QuantConnect account rather than only generate strategy code locally.

_Full upstream README: https://allmcps.com/mcp/quantconnect-mcp-server/readme_