Tidy Index delivers clean, well-organized datasets through a single API designed for humans, machines, and LLM agents alike.
for agents
GET /llms.txt →
a machine-readable invitation. No form, no salesperson.
§ 01 · one dataset, three shapes
Below is the irs-990 dataset — the annual return
every U.S. nonprofit files with the IRS, normally distributed as a
sprawling PDF or wall of e-file XML. Tidy Index serves it three ways:
a single record for your API, a streaming bulk export for your data
team, or a token-efficient summary your agents can drop straight into
context.
irs-990
{
"ein": "20-0049703",
"name": "Wikimedia Foundation, Inc.",
"fiscal_year": 2023,
"revenue_usd": 180249000,
"expenses_usd": 169106000,
"net_assets_usd": 254971000,
"source": "https://apps.irs.gov/app/eos/details/?ein=200049703"
}
{"ein":"20-0049703","name":"Wikimedia Foundation, Inc.","revenue_usd":180249000}
{"ein":"20-0097189","name":"Mozilla Foundation","revenue_usd":34019000}
{"ein":"26-1544963","name":"Khan Academy, Inc.","revenue_usd":85436000}
…
Wikimedia Foundation, Inc. (EIN 20-0049703) reported
$180.25M in revenue against $169.11M in expenses on
its 2023 Form 990, ending the year with $254.97M in
net assets.
Source: https://apps.irs.gov/app/eos/details/?ein=200049703
§ 02 · what it is
Some of what we serve comes from the open web. The rest is licensed from publishers, data vendors, and partners who’ve trusted us with their private feeds. All of it lives behind the same API, with the same stable schemas, the same stable IDs, and the same freshness guarantees — and one key gets you every record in every shape your tools (or your agents) prefer.
§ 03 · what we cover
A small slice of what’s already in the catalog.
100+ datasets — and the next one we add is whichever one you ask for.
§ 04 · start