For developers · Model Context Protocol
Use the Forage Calendar data in your AI assistant
Forage Calendar publishes an open-source MCP server that lets an AI assistant like Claude answer foraging questions from this site's vetted dataset instead of guessing. Grounding the model in real, cited data is what keeps it from confidently inventing a species that is out of season, out of range, or, worse, toxic.
What is MCP?
The Model Context Protocol is an open standard for giving an AI model live access to tools and data. Point a compatible client (such as Claude Desktop) at our server and the assistant can look things up in the Forage Calendar dataset mid-conversation.
What it can do
whats_in_seasonWild edibles in peak or shoulder season for any of our 52 US ecoregions in a given month. It will not list species that are out of range or out of season.
species_infoA vetted profile for a species: habitat and edibility as verbatim cited quotes, seasonality by region, and harvest cautions.
lookalikesThe documented toxic or confusable look-alikes for an edible, and how to tell them apart, drawn from our curated look-alike profiles.
list_regionsThe 52 ecoregions and their IDs, so the assistant can resolve a place to a region it can query.
Get it
The server, the dataset snapshot, and an evaluation that measures how much grounding improves the model's answers are all open source:
Safety note. An AI assistant, grounded or not, is a reference, not a field guide. Never eat any wild plant or fungus you have not identified with absolute certainty. Many edible species have toxic look-alikes.
Dataset compiled from occurrence data published via GBIF and iNaturalist, licensed CC BY 4.0.