> ## Documentation Index
> Fetch the complete documentation index at: https://docs.useforecast.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Overview

# API Reference

Forecast should document endpoints in two layers:

<Columns cols={2}>
  <Card title="Narrative API pages" icon="square-terminal">
    Handcrafted endpoint pages explain intent, caveats, common query patterns, and real request or response examples.
  </Card>

  <Card title="Schema-backed reference" icon="file-json">
    OpenAPI provides the canonical machine-readable contract for generated references, SDKs, and future testing surfaces.
  </Card>
</Columns>

## What makes a strong endpoint page

* A short introduction that explains when to use the endpoint
* One request pinned in the sidebar
* One response pinned in the sidebar
* Typed request fields
* Typed response fields
* Multi-language examples for the most common path
* Clear notes about what is live today versus still planned

## Current reference groups

<Columns cols={2}>
  <Card title="Realtime Data" icon="radio-tower" href="/realtime/markets">
    Market lists, tick windows, and latest order-book snapshots.
  </Card>

  <Card title="Product APIs" icon="workflow" href="/product/backtests">
    User backtests, research runs, bots, and strategy objects.
  </Card>
</Columns>

## Schema strategy

Right now the best pattern for Forecast is:

1. Use manual MDX endpoint pages for the high-value public surfaces.
2. Keep the OpenAPI file alongside them as the contract seed.
3. Expand the schema over time until autogenerated API sections become worthwhile.