Axiomatic MCP Servers

MCP (Model Context Protocol) servers that provide AI assistants with access to the Axiomatic_AI Platform - a suite of advanced tools for scientific computing and document processing.

🚀 Quickstart

1. Check system requirements

• Python
  • Install here
• uv
  • Install here
  • Recommended not to install in conda (see Troubleshooting)

2. Install your favourite client

Cursor installation

3. Get an API key

You will receive an API key by email shortly after filling the form. Check your spam folder if it doesn't arrive.

4. Install Axiomatic Operators

claude mcp add axiomatic-mcp --env AXIOMATIC_API_KEY=your-api-key-here -- uvx --from axiomatic-mcp all

1. Open Claude Desktop settings → Developer → Edit MCP config
2. Add this configuration:

{
  "mcpServers": {
    "axiomatic-mcp": {
      "command": "uvx",
      "args": ["--from", "axiomatic-mcp", "all"],
      "env": {
        "AXIOMATIC_API_KEY": "your-api-key-here"
      }
    }
  }
}

1. Restart Claude Desktop

Follow the MCP install guide and use the standard configuration above. See the official instructions here: Gemini CLI MCP Server Guide

{
  "axiomatic-mcp": {
    "command": "uvx",
    "args": ["--from", "axiomatic-mcp", "all"],
    "env": {
      "AXIOMATIC_API_KEY": "your-api-key-here"
    }
  }
}

Follow the Windsurf MCP documentation. Use the standard configuration above.

{
  "axiomatic-mcp": {
    "command": "uvx",
    "args": ["--from", "axiomatic-mcp", "all"],
    "env": {
      "AXIOMATIC_API_KEY": "your-api-key-here"
    }
  }
}

Click the button to install:

Note: After installing via the button, open LM Studio MCP settings and add:

"env": {
  "AXIOMATIC_API_KEY": "your-api-key-here"
}

Create or edit the configuration file ~/.codex/config.toml and add:

[mcp_servers.axiomatic-mcp]
command = "uvx"
args = ["--from", "axiomatic-mcp", "all"]
env = { AXIOMATIC_API_KEY = "your-api-key-here" }

For more information, see the Codex MCP documentation

Use this server configuration:

{
  "command": "uvx",
  "args": ["--from", "axiomatic-mcp", "all"],
  "env": {
    "AXIOMATIC_API_KEY": "your-api-key-here"
  }
}

Note: This installs all tools under one server. If you experience other issues, try individual servers instead.

Reporting Bugs

Found a bug? Please help us fix it by creating a bug report.

Connect on Discord

Join our Discord to engage with other engineers and scientists using Axiomatic Operators. Ask for help, discuss bugs and features, and become a part of the Axiomatic community!

Troubleshooting

Cannot install in Conda environment

It's not recommended to install axiomatic operators inside a conda environment. uv handles seperate python environments so it is safe to run "globally" without affecting your existing Python environments

Server not appearing in Cursor

1. Restart Cursor after updating MCP settings
2. Check the Output panel (View → Output → MCP) for errors
3. Verify the command path is correct

The "Add to cursor" button does not work

We have seen reports of the cursor window not opening correctly. If this happens you may manually add to cursor by:

1. Open cursor
2. Go to "Settings" > "Cursor Settings" > "MCP & Integration"
3. Click "New MCP Server"
4. Add the following configuration:

{
  "mcpServers": {
    "axiomatic-mcp": {
      "command": "uvx --from axiomatic-mcp all",
      "env": {
        "AXIOMATIC_API_KEY": "YOUR API KEY"
      },
      "args": []
    }
  }
}

Multiple servers overwhelming the LLM

Install only the domain servers you need. Each server runs independently, so you can add/remove them as needed.

API connection errors

1. Verify your API key is set correctly
2. Check internet connection

Tools not appearing

If you experience any issues such as tools not appearing, it may be that you are using an old version and need to clear uv's cache to update it.

uv cache clean

Then restart your MCP client (e.g. restart Cursor).

This clears the uv cache and forces fresh downloads of packages on the next run.

Individual servers

You may find more information about each server and how to install them individually in their own READMEs.

🖌️ AxEquationExplorer

Compose equation of your interest based on information in the scientific paper.

📄 AxDocumentParser

Convert PDF documents to markdown with advanced OCR and layout understanding.

📝 AxDocumentAnnotator

Create intelligent annotations for PDF documents with contextual analysis, equation extraction, and parameter identification.

📊 AxPlotToData

Extract numerical data from plot images for analysis and reproduction.

⚙️ AxModelFitter

Fit parametric models or digital twins to observational data. Describe the model and data in plain language — the server generates executable JAX fitting code and runs it in a sandboxed environment.

⚙️ AxModelFitter (Legacy)

Deprecated — superseded by AxModelFitter above; will be removed in the next major release.

🧮 AxArgmin

Numerical optimization, rootfinding, ODE simulation, and optimal control. Describe the problem in plain language — the server generates and runs the corresponding code in a sandboxed environment.

📚 AxKnowledgeBase

Semantic search over Axiomatic's curated Knowledge Base — scientific papers, entities, and passages, always returned with their source for citation. Also exposes your organization's own private knowledge graph: ingest a PDF into it, then search and query it with the same tools. Both graphs answer read-only Cypher for when the answer has to be a table.

📐 AxBlueprints

Browse Axiomatic's component blueprints and read their contracts — ports, modes, parameters, lifecycle status, and the validity, source and verification notes — one section at a time.

🔎 AxPaperSearch

Search arXiv and OpenAlex for scientific papers — abstracts, authors, DOIs, citation counts, and direct PDF links.

🌊 AxTidy3D

Generate and run Tidy3D electromagnetic simulations (FDTD, mode solving) from natural language, with a cost-safe estimate-then-confirm flow for cloud runs.

💡 AxMeep

Generate and run Meep FDTD simulations from natural language. Simulations run as remote jobs; figures come back as inline images and arrays are summarized and saved locally. Requires an API key with playground access.

Requesting Features

Have an idea for a new feature? We'd love to hear it! Submit a feature request and:

• Describe the problem your feature would solve
• Explain your proposed solution
• Share any alternatives you've considered
• Provide specific use cases

Support

• Join our Discord Server
• Issues: GitHub Issues