Summary

Bulk lookup is what makes it useful beyond one-off questions.

Single-food nutrition answers are easy to get anywhere. What this adds is `get_multiple_foods` and nutrient filtering, so a comparison across a dozen ingredients is one round trip with only the columns you asked for — which is the difference between an assistant that can plan a menu and one that can recite a label.

What it is

A Python server over the USDA FoodData Central API. It covers the Foundation, SR Legacy, Survey and Branded datasets, so answers come from the USDA's own records rather than from the model's recollection of a nutrition label.

What you get

`search_foods` searches the database, taking `query`, `page_size` and `page_number`
`get_food_details` returns the full record for one `fdc_id`, optionally narrowed to specific `nutrients`
`get_multiple_foods` looks up a list of `fdc_ids` in one call
`get_food_nutrients` pulls just the nutrients you name across several foods
`list_foods` browses by `data_type` — Foundation, SR Legacy, Survey or Branded

Requirements

A free USDA API key from api.nal.usda.gov, set as `USDA_API_KEY`. Python 3.11+ with `uv sync`, then `uv run main.py`; the project is `usda-api-mcp`, version 2.1.0. There is also a macOS GUI installer (macOS 10.15+) that writes the client config for you. The free tier allows 3600 requests/hour.

USDA api