> ## 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.

# MCP server

> Give Claude, ChatGPT, Claude Code, Codex or Cursor live access to the API.

> 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.

```text theme={null}
https://data.hyperdash.com/mcp
```

The remote MCP server uses Streamable HTTP. It has one tool per endpoint that your connection's scopes allow, plus `time_window` to convert local dates and times, and `create_job`, `get_job` and `get_job_result` for async jobs (Growth and up).

There are two ways to sign in:

* **OAuth:** you sign in to the console and approve one project and its scopes. No key is involved, and you can revoke access from the console at any time. An OAuth connection can call every data tool its scopes allow. It cannot open streams or read your credit balance: use an API key for those.
* **API key:** the client sends `Authorization: Bearer hdk_sk_...` from its own configuration. Never put a key in a prompt.

## Set up your agent

The console's **MCP** page shows these steps for each client, with a button to copy the setup.

> [tab] Claude
>
> 1. Open **Settings**, then **Connectors**, and select **Add custom connector**.
> 2. Name it **Hyperdash Data** and paste `https://data.hyperdash.com/mcp`.
> 3. Select **Connect** and sign in with your Hyperdash account.
> 4. Choose the project and the scopes, then select **Approve access**.
>
> The connector is saved to your Claude account, so it works on web, desktop and mobile.

> [tab] ChatGPT
>
> 1. Open **Plugins**, select **+**, then **Add custom MCP server**.
> 2. Paste `https://data.hyperdash.com/mcp` and choose OAuth authentication.
> 3. Sign in with your Hyperdash account.
> 4. Choose the project and the scopes, then select **Approve access**.

> [tab] Claude Code
>
> Run this once in your shell, with your key saved as `HYPERDASH_API_KEY`:
>
> ```bash wrap theme={null}
> claude mcp add --transport http --scope user hyperdash-data https://data.hyperdash.com/mcp --header "Authorization: Bearer $HYPERDASH_API_KEY"
> ```
>
> To sign in with OAuth instead, leave out `--header`, then run `/mcp` in Claude Code, pick `hyperdash-data` and choose **Authenticate**.

> [tab] Codex
>
> Run this once in your shell, with your key saved as `HYPERDASH_API_KEY`:
>
> ```bash wrap theme={null}
> codex mcp add hyperdash-data --url https://data.hyperdash.com/mcp --bearer-token-env-var HYPERDASH_API_KEY
> ```
>
> To sign in with OAuth instead, leave out `--bearer-token-env-var HYPERDASH_API_KEY`, then run `codex mcp login hyperdash-data`.

> [tab] Cursor
>
> Add this to `~/.cursor/mcp.json`, or to `.cursor/mcp.json` in one project:
>
> ```json theme={null}
> {
>   "mcpServers": {
>     "hyperdash-data": {
>       "url": "https://data.hyperdash.com/mcp",
>       "headers": {
>         "Authorization": "Bearer ${env:HYPERDASH_API_KEY}"
>       }
>     }
>   }
> }
> ```
>
> To sign in with OAuth instead, leave out `headers`, then open **Cursor Settings**, then **MCP**, and select **Connect** next to `hyperdash-data`.

> [tab] Other
>
> Add a remote MCP server with the URL `https://data.hyperdash.com/mcp`. If the client supports MCP OAuth, sign in when it opens the console. Otherwise set the `Authorization: Bearer` header in the client's configuration.

> [!NOTE]
>
> The **MCP** entry in this page's menu installs a different server, `docs.hyperdash.com/mcp`, which searches these docs. The data server is `data.hyperdash.com/mcp`.

## Try it

Open a new chat and ask:

```text wrap theme={null}
Which markets had the most liquidations in the last 24 hours?
```

For better results, have the agent read the [Data skill](https://docs.hyperdash.com/skills/data) first.

## Spending limits

Every MCP call uses your workspace's credits. A connection has no budget unless you set one, so by default an agent can spend up to your balance. MCP results do not show what a call cost; the console's **Logs** page shows the credits each request used. To keep an agent inside a budget, open **Settings**, then **Security**, then **Connected apps** in the console, edit the connection and set a credit budget per day or month.

A call over any limit is refused without charge, and the error says which limit it hit. See [Credits](https://docs.hyperdash.com/data-api/credits.md#controlling-spend).


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