Origin Mcp

origin-mcp

[](https://pypi.org/project/origin-mcp/) [](https://pepy.tech/projects/origin-mcp) [](https://pypi.org/project/origin-mcp/) [](LICENSE)

[简体中文](README.zh.md)

origin-mcp is a local Model Context Protocol (MCP) server that lets AI assistants control Origin/OriginPro on Windows. It connects through OriginLab's Python automation interface and exposes tools for importing data, editing worksheets, creating and refining graphs, running Origin analyses, exporting figures, and managing the Origin application lifecycle.

This project is still in a testing stage. Trying it on real Origin workflows, reporting issues, suggesting improvements, and opening pull requests are all welcome.

Highlights

• Import CSV, TSV, TXT, DAT, XLS, and Excel data into Origin worksheets.
• Read, write, sort, clear, and export worksheet data.
• Create and refine common 2D, 3D, contour, statistical, and specialized plots.
• Run Origin analyses such as fitting, smoothing, integration, peak finding,

descriptive statistics, interpolation, normalization, t-tests, FFT/IFFT, and correlation.

• Export figures and projects through a local Origin GUI bridge.
• Save a finished graph as a reusable user template, then search/match and reapply

it to same-type figures (see [docs/tools.md](docs/tools.md#user-template-library)).

Nature-Style Figures

By default, origin-mcp keeps the styling defined by the Origin graph template you are using. If you want a cleaner publication-style scientific figure, ask your AI assistant to use the Nature-style format when creating or refining the graph. The preset applies colorblind-aware palettes, stronger scientific plot strokes, Arial typography, and simpler legends.

For more control, you can ask the assistant to list available palettes. See [docs/tools.md](docs/tools.md#palette-catalog) for the detailed palette and style controls.

Requirements

• Windows
• Origin or OriginPro installed and licensed
• Origin/OriginPro 2026 is the primary tested target; other Origin versions are

not currently guaranteed

• Origin's embedded Python with the preinstalled originpro package

Python version support

origin-mcp runs as two cooperating processes, and the supported Python versions differ by role:

• MCP server core (the python -m origin_mcp process, which only talks to

the bridge over localhost): Python 3.10+. CI tests this core on Windows with Python 3.10, 3.11, 3.12, 3.13, and 3.14.

• Origin bridge (addon.py): runs inside Origin's own embedded Python, so

its version is whatever your Origin install ships — there is nothing to pick.

Direct external originpro automation is not a supported MCP backend for this project. Start the bridge inside Origin's embedded Python and let the MCP server connect to it over localhost.

Installation

Install the MCP server core from PyPI:

pip install origin-mcp

That is all the MCP server needs: it runs as python -m origin_mcp and reaches Origin only through the bridge over localhost. The bridge runs inside Origin's own embedded Python and installs its own dependencies (see [Start the Origin Bridge](#start-the-origin-bridge)).

An optional origin-mcp[origin] extra pulls originpro and pywin32 into the same environment; the standard bridge setup does not require it. To work from a checkout instead, run pip install -e . in the repository root.

Agentic Setup

Copy this to your AI agent and let it self-configure:

Fetch and follow this bootstrap guide end to end:
https://raw.githubusercontent.com/Ge-Shun/origin-mcp/main/docs/agentic/origin-mcp-bootstrap.md

MCP Configuration

Example MCP client configuration:

{
  "mcpServers": {
    "origin": {
      "command": "python",
      "args": ["-m", "origin_mcp"]
    }
  }
}

If python is not the Python 3.10+ interpreter you installed origin-mcp into, use that interpreter's absolute python.exe path instead. More examples are in [docs/mcp-config.md](docs/mcp-config.md).

Start the Origin Bridge

The bridge runs inside Origin's own Python so originpro stays on Origin's UI thread. There is nothing to configure — start it once per Origin session:

Origin Apps (recommended for daily use). Build and install the two bridge Apps once with the steps in [docs/origin-ui-buttons.md](docs/origin-ui-buttons.md). After that, click Origin MCP Bridge Start in the Apps gallery to start the bridge and Origin MCP Bridge Stop to stop it.

Python Console (one-off or troubleshooting). Open Origin's Python Console and paste this line (replace the path with your checkout):

import runpy; runpy.run_path(r"C:\path\to\origin-mcp\addon.py", run_name="__main__")

A Bridge is running inside Origin. box confirms startup; keep that console running while you use the tools. To stop, ask your MCP assistant to shut the bridge down (it calls origin_bridge_shutdown), or double-click scripts\stop-bridge.cmd (or run python scripts\stop_bridge.py). Origin stays open either way.

If a package is missing or the bridge will not start, see [docs/origin-bridge.md](docs/origin-bridge.md).

Security

The bridge listens only on 127.0.0.1 and authenticates local requests by default with a per-session token, so normal use needs no security setup.

Treat that token as a credential. Any local process that presents it can drive Origin with the full tool surface, including arbitrary LabTalk execution through origin_run_labtalk. The token is generated per session and written to an owner-scoped file in your per-user temporary directory (%TEMP%/origin-mcp/bridge.json on Windows), which a standard single-user machine already protects through the directory's OS permissions. If you redirect TEMP or ORIGIN_MCP_BRIDGE_HANDSHAKE to a directory other local users can read, the token — and therefore control of Origin — is exposed to them. Setting ORIGIN_MCP_BRIDGE_NO_AUTH removes the token boundary entirely and should be used only when you fully trust every local process.

If you need to restrict which files tools may read or write, set ORIGIN_MCP_ALLOWED_ROOTS to the allowed directories. Avoid disabling bridge authentication unless you fully trust every local process on the machine.

License

MIT. See [LICENSE](LICENSE).

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: Ge-Shun
• Source: Ge-Shun/origin-mcp
• License: MIT

Install and usage instructions live in the source repository linked above.