Jij

You run the Jij MCP Server to access tools that support mathematical optimization with JijModeling and to run quantum computing tasks via Qiskit. It unifies modeling assistance, code checking, and quantum workflow support in a single MCP environment, so you can build optimization models and experiment with quantum workflows from a single place.

How to use

You connect to the Jij MCP Server using an MCP client. You can start the server in one of two ways: by running it locally with the MCP client’s stdio mode or by launching it through Docker.

How to use

• Start a local (stdio) session with the MCP client: you run the exact command you specify in your configuration to execute the server script. This starts the Python-based Jij MCP Server and exposes its tools to your client.

How to install

Prerequisites: ensure you have a working Python environment and the MCP client tooling installed on your system. You will also need network or local access to the server so you can run either a local stdio session or a Dockerized instance.

How to install

Step 1: Clone the Jij MCP Server repository using your preferred method for this project. Step 2: Install the required dependencies for the server. Step 3: Configure the server in your MCP client by using one of the provided run configurations. Step 4: Start the server using one of the available start commands.

Configuration and usage notes

The Jij MCP Server provides two explicit runtime configurations you can use to start the server from your MCP client. You can choose the stdio-based approach for a local run or the Docker-based approach for a containerized run.

Configuration and usage notes

Stdio (local) configuration starts the Python server script jij_mcp/server.py using the MCP client runtime. Example usage is shown in the configuration snippet below. You can adapt the <YOUR PATH> placeholder to point to your local copy of the server.

Configuration and usage notes

Docker (container) configuration runs the published image ghcr.io/jij-inc/jij-mcp-server:latest via Docker. This provides an isolated environment and easy startup. Replace the placeholder with your actual environment setup.

Notes

No explicit client-side commands or protocols are required beyond starting the server with your MCP client in either stdio or Docker mode. Ensure the server script and Docker image are accessible from your environment, and configure the path and server script name as shown in the examples.

Security considerations

Keep access to the server restricted to trusted clients. If you expose the server over a network, apply appropriate authentication and encryption as you would for any service that handles optimization models and quantum tooling.

Troubleshooting tips

If the server fails to start, verify that the specified directory or image is accessible, that the server script jij_mcp/server.py exists, and that the MCP client is correctly configured to launch the command. Check logs for any errors related to dependencies or environment configuration.

Available tools

learn_jijmodeling

Guide to JijModeling syntax and usage

jm_check

Validation tool for JijModeling code

qiskit_v0tov1v2_migration_guide

Migration guide for transitioning between Qiskit versions

qiskit_v1_api_reference_toc

Table of contents for Qiskit v1 API reference

qiskit_v2_api_reference_toc

Table of contents for Qiskit v2 API reference

qiskit_tutorial

Access to IBM Quantum Learning Hub tutorials