> ## Documentation Index
> Fetch the complete documentation index at: https://docs.0xarchive.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Core concepts

> Understand the choices that determine API namespace, market availability, freshness, limits, credits, and data rights.

Make a few choices before you open the endpoint reference: the venue and market type, the data type and depth, the time window, how the data reaches you, and the checks the result must pass. Each concept below settles one of those choices.

## Concepts and what they decide

| Concept | What it decides |
| - | - |
| [Venues and market types](/core-concepts/venue-taxonomy) | Whether a request belongs to Hyperliquid core, Spot, HIP-3, HIP-4, or Lighter, and which namespace and symbol format it uses |
| [Venue coverage](/venue-coverage) | Whether the exact market and data type are available, and from when |
| [Order book depth](/core-concepts/order-book-depth) | Whether the job needs L2, L3, L4, diffs, order events, or history |
| [Plan a historical pull](/core-concepts/historical-market-data) | Whether the job needs snapshots, history, replay, exports, or backtesting metadata |
| [Replay and reconstruction](/core-concepts/replay-and-reconstruction) | Whether event order, local state, or gap handling matters |
| [API keys](/authentication) | How REST, WebSocket, SDK, CLI, MCP, and Agent skill clients pass credentials |
| [Responses](/responses) and [Errors](/errors) | What to read and log from success and error responses |
| [Rate limits](/rate-limits) | How fast and how broadly a client can call the API |
| [Credits](/core-concepts/credits) | What a request costs and how export credits apply |
| [Data quality](/data-quality) | Whether a window is fresh and complete enough, and how to treat gaps |
| [Data rights](/data-rights) | What a downstream product can do with API output |

## Choose the venue before the endpoint

Choose the venue and market type before the endpoint. Hyperliquid includes core perpetuals, Spot, HIP-3, and HIP-4. Lighter uses its own API namespace. Do not route a Spot pair through a core perpetual endpoint, strip a HIP-3 builder prefix, or treat HIP-4 probability-like prices as ordinary dollar prices.

Then choose the data type: order book, trades, candles, funding, open interest, liquidations, orders, L3, L4, replay, or data quality. Check whether the exact market and time window are covered. Size the request against account capacity and request-window limits separately.

## From concepts to a first request

<Steps>
  <Step title="Pick the venue and market type">
    Choose Hyperliquid or Lighter, then the market type that matches the market identity.
  </Step>

  <Step title="Confirm availability">
    Use instrument, coverage, or freshness routes before widening to long windows or many symbols.
  </Step>

  <Step title="Choose the interface">
    Use the REST API for request and response jobs, WebSocket for real-time channels or channel-specific replay, the SDKs for application code, the CLI for shell jobs, the MCP server for MCP-capable hosts, and the Agent skill for reusable agent workflows.
  </Step>

  <Step title="Save trace fields">
    Store the route, parameters, time window, data-quality result, and any request ID the route returns with every output that another system will trust.
  </Step>
</Steps>

Write these down before broad work starts: the depth and state model the job needs, its capacity rule (request rate, concurrency, credits, standard replay speed, request-window limits, and stop conditions), and the trace fields it keeps (route, parameters, quality decision, cursor, and request ID). When you move to code, the [REST API](/rest-api), [Endpoint reference](/api-reference), [OpenAPI](/openapi), and [WebSocket](/websocket) pages hold the exact contracts.

## Next step

Pick the namespace and symbol format for your market in [Venues and market types](/core-concepts/venue-taxonomy).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.