Setup (once)

Run the MCP server locally:

bash

uv run python src/essdive_mcp/main.py --token YOUR_ESS_DIVE_TOKEN_HERE

Or provide the token via a file:

bash

uv run python src/essdive_mcp/main.py --token-file /path/to/token.txt

Note: the environment variable name is ESSDIVE_API_TOKEN (no underscore between ESS and DIVE).

bash

claude mcp add --transport stdio essdive-mcp --env ESSDIVE_API_TOKEN=YOUR_ESS_DIVE_TOKEN_HERE -- uv run python ./src/essdive_mcp/main.py

Token file alternative:

bash

claude mcp add --transport stdio essdive-mcp -- uv run python ./src/essdive_mcp/main.py --token-file /path/to/token.txt

Tools

Search for fields named "temperature" with at least 100 records:

code

search-ess-deepdive with field_name="temperature" and record_count_min=100

Filter by DOI and field definition text:

code

search-ess-deepdive with doi="10.15485/2453885" and field_definition="soil" and page_size=25

Get detailed field metadata for a dataset file:

code

get-ess-deepdive-dataset with doi="10.15485/2453885" and file_path="dataset.zip/data.csv"

Get full file metadata and download info:

code

get-ess-deepdive-file with doi="doi:10.15485/2453885" and file_path="dataset.zip/data.csv"

Create map links for a point (geojson.io + Google Maps + Google Earth KML):

code

coords-to-map-links with points=[[38.9219, -106.9490]] and zoom=12