eo-processor-mcp

An MCP (Model Context Protocol) server for Earth Observation processing, powered by eo-processor.

Similar to stac-mcp for STAC catalog operations, this server exposes eo-processor's high-performance Rust-accelerated EO computation functions as MCP tools that AI agents can invoke.

Features

• 15 MCP tools covering spectral indices, change detection, temporal statistics, masking, morphology, distances, trend analysis, classification, texture features, zonal statistics, and more
• Rust-accelerated computation via eo-processor's PyO3 bindings (GIL-released, multi-core parallel)
• stdio and HTTP transport support
• Observability built in (structured logging, metrics, correlation IDs, latency histograms)
• Prompts to guide AI agents in using the tools effectively

Tools

Installation

pip install -e .
# or with uv
uv pip install -e .

Usage

stdio (default)

eo-processor-mcp
# or
python -m eo_processor_mcp

HTTP/SSE

EO_PROCESSOR_MCP_TRANSPORT=http python -m eo_processor_mcp

MCP Client Configuration

{
  "eo-processor": {
    "command": "uvx",
    "args": ["--from", "git+https://github.com/BnJam/eo-processor-mcp", "eo-processor-mcp"],
    "transport": "stdio"
  }
}

Data Format

All tools accept NumPy .npy file paths as input and write results to .npy files. Tool responses include:

• output_path: Path to the result .npy file
• stats: Summary statistics (shape, dtype, min, max, mean, std, NaN count)

Environment Variables

Development

make install-dev    # Install with dev dependencies
make test           # Run tests
make lint           # Run ruff
make format         # Auto-format
make coverage       # Coverage report

License

Apache-2.0