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

# AI agents

> Connect an AI agent to Hyperdash Data to ask questions or build an app.

> AI agents: read [https://docs.hyperdash.com/skills/data](https://docs.hyperdash.com/skills/data) before using this API. Endpoint index: [https://data.hyperdash.com/llms.txt](https://data.hyperdash.com/llms.txt). Never ask the user to paste an API key into the chat.

**1. Give your agent access**

Pick where your agent runs.

> [tab] Claude or ChatGPT
>
> No API key is needed: you sign in with your Hyperdash account.
>
> 1. In Claude, open **Settings**, then **Connectors**, and select **Add custom connector**. In ChatGPT, open **Plugins**, select **+**, then **Add custom MCP server**.
> 2. Paste `https://data.hyperdash.com/mcp`.
> 3. Sign in with your Hyperdash account, choose the project and scopes, and select **Approve access**.

> [tab] Coding agents
>
> For Claude Code, Codex, Cursor and other agents that run on your computer.
>
> 1. Create a key in the [console](https://console.hyperdash.com) under **API Keys**. Copy it right away: it is shown only once.
> 2. Set it as an environment variable in the window where you start your agent:
>
>    ```bash theme={null}
>    export HYPERDASH_API_KEY="hdk_sk_..."
>    ```
>
>    The agent reads the key from the environment, so you never paste it into the chat. To keep the key after you close the window, add the same line to your shell profile, such as `~/.zshrc`.
>
> The master prompt in the next step has the agent connect itself.

**2. Send the master prompt**

This prompt has your agent connect, read the Data skill and summarize what it can do for you.

The master prompt, for the user to send to their agent:

```text theme={null}
Set up Hyperdash Data, the Hyperliquid data API, and learn what it offers.

1. If the Hyperdash Data MCP server is not connected yet, connect it at https://data.hyperdash.com/mcp (Streamable HTTP). Sign me in with OAuth if you can. Otherwise my key is in the HYPERDASH_API_KEY environment variable: send it as an "Authorization: Bearer" header and never print it. If you cannot add the server yourself, tell me the steps.
2. Read https://docs.hyperdash.com/skills/data and follow it from now on.
3. If you are not connected over MCP, or you are going to write code against the API, read https://data.hyperdash.com/llms.txt. It indexes every data endpoint. Look up parameters the way the skill describes.
4. If this project already has an instructions file (CLAUDE.md or AGENTS.md), add a short note with both links so you remember them next session. Do not create one.
5. Then tell me what you can now do for me, grouped by markets, wallets, liquidations, trader analytics, exchange-wide data and live streams, and suggest five questions worth asking first.
```

**3. Try it**

Paste one of these into your agent.

```text wrap theme={null}
Where are the largest clusters of ETH liquidation levels within 5% of the current price?
```

```text wrap theme={null}
Which wallets opened the largest BTC positions in the last 24 hours, and how has each one traded over the past 90 days?
```

```text wrap theme={null}
Which markets are profitable traders most one-sided on right now?
```

```text wrap theme={null}
Which markets had the most liquidations in the last 24 hours, and which hour was the worst?
```

[Recipes](https://docs.hyperdash.com/data-api/recipes.md) has more prompts.

**4. Ask for a full integration**

Without further direction, a coding agent often wires market prices and stops. This prompt points it at the Build skill, which has it wire all four levels (market data, wallet intelligence, live streams and account status) and report what it left out.

```text wrap theme={null}
Read https://docs.hyperdash.com/skills/build and follow it. Integrate Hyperdash Data into this app as a full integration: market data, wallet intelligence, live streams and account status. My key is in the HYPERDASH_API_KEY environment variable. When you finish, tell me which levels you wired, which you left out and why.
```

## Skills

The prompts above point your agent at skills: files that teach an agent how to do a job with Hyperdash Data. The [Skill library](https://docs.hyperdash.com/data-api/skills.md) lists them, with a starter prompt for each and the install command.

## Which tool to use

| You want to | Use |
| - | - |
| Ask questions about Hyperliquid in a chat and get live answers | [MCP server](https://docs.hyperdash.com/data-api/mcp.md) |
| Have a coding agent build an app, bot or dashboard on the API | [Full integration prompt](#ask-for-a-full-integration) |
| Give an agent the API reference without installing anything | [Docs for agents](#docs-for-agents) |
| Limit what an agent can spend | [Spending limits](https://docs.hyperdash.com/data-api/mcp.md#spending-limits) |

## Docs for agents

Use these when an agent writes code against the API. All of them are free to read and need no key.

* **Data skill:** `https://docs.hyperdash.com/skills/data`. The entry point to the [Skill library](https://docs.hyperdash.com/data-api/skills.md). It redirects to `/.well-known/skills/hyperdash-data/SKILL.md`, the standard path that agents and install tools check.
* **Endpoint index:** `https://data.hyperdash.com/llms.txt`. Every data endpoint with a one-line description, its scope and its page size.
* **OpenAPI spec:** `https://data.hyperdash.com/openapi.json`. Every parameter of every endpoint. It is about 1.4 MB, so have the agent look up one endpoint at a time instead of loading it whole.
* **Docs index:** `https://docs.hyperdash.com/llms.txt`. Every page in these docs. Add `.md` to any page URL for a Markdown copy.

To make an agent remember the API across sessions, paste this into the instructions file it reads: `CLAUDE.md` for Claude Code, `AGENTS.md` for Codex and other agents, or the project instructions of a Claude Project.

```markdown wrap theme={null}
## Hyperdash Data API

- Skill: https://docs.hyperdash.com/skills/data. Read it before the first request.
- Endpoints: https://data.hyperdash.com/llms.txt. Use only endpoints listed there, plus the free `GET /v1/me`, `/v1/quote`, `/v1/status`, `/v1/datasets` and `/v1/changelog`.
- The key is in the `HYPERDASH_API_KEY` environment variable. Never print it, log it or commit it.
```

## Supported agents

| Agent | MCP server | Skills |
| - | - | - |
| Claude (web, desktop, mobile) | OAuth | Paste the link |
| ChatGPT | OAuth | Paste the link |
| Claude Code | OAuth or API key | Install or paste the link |
| Codex | OAuth or API key | Install or paste the link |
| Cursor | OAuth or API key | Install or paste the link |

Any other agent that supports remote MCP servers or the Agent Skills format works the same way.


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