eve-esa/mcp-tool-registry

About

The Model Context Protocol (MCP) is an open standard that lets an AI agent discover and call external tools. This repository is the tool registry of EVE, the Earth Virtual Expert: a collection of MCP servers for Earth Observation (EO) and Earth Science, written by the EVE team and the community.

• One folder per server. Each server in servers/ is self-contained: a FastMCP entrypoint, pinned requirements and optional tests.
• Checked on every pull request. CI validates the structure, lints the code, scans for secrets and builds the container.
• Deployed to EVE. Merged servers are deployed to AWS Bedrock AgentCore and become tools for the EVE agent.

Anyone building EO agents can run these servers locally or contribute a new one.

Servers

Quick start

Requires Python 3.13. Each server has its own requirements, so install only the one you need:

git clone https://github.com/eve-esa/mcp-tool-registry.git
cd mcp-tool-registry
python3 -m venv .venv && source .venv/bin/activate
pip install -r servers/geocode/requirements.txt

Servers that need credentials ship a .env.template: copy it to .env in the same folder and fill it in. Some also accept them per request as HTTP headers, see Per-request credential headers.

Run over HTTP (streamable HTTP on http://localhost:8000/mcp, the default):

python servers/geocode/server.py

Use it from an MCP client (stdio), for example Claude Desktop, Cursor or VS Code:

{
  "mcpServers": {
    "geocode": {
      "command": "/absolute/path/to/mcp-tool-registry/.venv/bin/python",
      "args": ["/absolute/path/to/mcp-tool-registry/servers/geocode/server.py", "--transport", "stdio"]
    }
  }
}

The MCP Inspector (npx @modelcontextprotocol/inspector) is the quickest way to list and call the tools of a running server. For curl examples and Docker builds, see Local testing.

Contributing

New EO tools are welcome. In short:

1. Create servers// with a FastMCP server.py and a pinned requirements.txt.
2. Add a CITATION.cff with your name, affiliation and GitHub handle. After merge, the server README is written from it.
3. Add a .env.template if the server needs secrets, and a test.py.
4. Run it locally and build it with docker build -f shared/Dockerfile servers//.
5. Open a pull request against main, one server per PR.

CI checks the folder structure, the entrypoint contract, pinned versions, a valid CITATION.cff, secrets and file sizes, then runs ruff and a Docker build. The full guide, with a server template and the PR checklist, is in CONTRIBUTING.md.

Deployment and early access

Servers merged into main are deployed automatically to AWS Bedrock AgentCore, where the EVE agent uses them. If you want early access to the hosted servers before they are public, write to [email protected].

Earth Virtual Expert (EVE)

Earth Virtual Expert (EVE) aims to advance the use of Large Language Models (LLMs) within the Earth Observation (EO) and Earth Science (ES) community.

• Website: https://eve.philab.esa.int/
• Hugging Face: https://huggingface.co/eve-esa
• Other repositories: https://github.com/eve-esa

Funding

This project is supported by the European Space Agency (ESA) Φ-lab through the Large Language Model for Earth Observation and Earth Science project, as part of the Foresight Element within FutureEO Block 4 programme.

Citation

If you use this project in academic or research settings, please cite:

@misc{atrio2026evedomainspecificllmframework,
      title={{EVE}: A Domain-Specific {LLM} Framework for Earth Intelligence},
      author={Àlex R. Atrio and Antonio Lopez and Jino Rohit and Yassine El Ouahidi and Marcello Politi and Vijayasri Iyer and Umar Jamil and Sébastien Bratières and Nicolas Longépé},
      year={2026},
      eprint={2604.13071},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2604.13071},
}

License

This project is released under the Apache 2.0 License. See the LICENSE file for details.