Congratulations to the winners of the EVE EO MCP tool hackathon
The EVE EO MCP tool hackathon was held as part of the Agentic AI for Earth Observation workshop in Berlin, in collaboration with BIFOLD (Berlin Institute for the Foundations of Learning and Data) and ESA Φ-lab. Teams spent the day turning Earth observation services into tools that EVE can use, then combining them into working workflows.
The jury judged each project on four things: how relevant it is to real Earth observation workflows, the technical quality of its integration with EVE, originality, and how reusable it is for the wider community.
1st place: OrbitCast
Built by Cristopher Castro Traba, OrbitCast tells you when the next satellite image of an area will be taken and how cloudy it will be.
2nd place: Solenix EO toolset
Built by Cesar Guzman and Matteo Zuracki, the toolset has three MCP servers and four use cases, from deforestation checks on cocoa farms to mudslide risk after a wildfire. Each use case has its own notebook with maps, tests and sources.
3rd place: EO Spatial Split Planner
Built by Stefano Francesco Pitton and Hugo Lee, this is a tool for real machine learning work on satellite images. It splits an area into training and test patches that are far enough apart, so a model is not tested on near-copies of its training data.
Thank you to everyone who took part, and to our jury: Àlex R. Atrio, Antonio Lopez and Emanuele Micieli from Pi School, and Nicolas Longépé, Eva Gmelich Meijling and Riccardo D’Ercole from ESA Φ-lab.
