The index as a data product
A read REST API over the same relational backbone the site runs on, every response carries its source, source_tier and confidence, so anything you build on it stays citable. Auth + base URL on launch.
Is the Fieldbook API live?
Not yet, it is in build. The endpoints listed above are the planned surface, and the base URL and authentication will be published at launch. The data behind them is already live and citable on the site.
What does each API response include?
The record, plus where it came from. Every row carries its source, that source's tier, and a confidence grade. Those are the same fields used internally, so anything you build stays traceable back to the evidence.
Which endpoints are planned?
The agent catalogue and single-agent spec sheets, the foundation-model list, the supply-chain graph, MCP adoption, and a universal entity search. All GET, all JSON.
How will the API be authenticated?
Auth and the base URL will be confirmed at launch. It is a read-only API, so access is scoped to retrieval and there is nothing to mutate.
Should I use the API, the CLI or the dataset?
They return the same records. Use the API for live lookups from an application, the CLI for quick queries in a terminal, and the dataset when you want the whole index locally.