Get started - MCP

Use agent-data’s remote MCP server to give AI agents access to real-time, structured data. Agents use it to discover endpoints, inspect docs, and call routes from Claude, Cursor, ChatGPT, or any other MCP client.

Installation

The server URL is the same for every client:

https://agent-data.motie.dev/mcp

To authenticate, sign in to your agent-data account through the browser window that opens.

The sections below cover common clients. Other clients need remote HTTP MCP support and OAuth sign-in to connect to this server.

Claude Code

Add agent-data as a remote MCP server:
Terminal window
```
claude mcp add --transport http agent-data https://agent-data.motie.dev/mcp
```
Add --scope user to make it available in every project, not just the current one.
Inside Claude Code, run /mcp, choose agent-data, select Authenticate, and complete the sign-in that opens in your browser.

The agent-data tools are available in your next session.

Troubleshooting: If agent-data shows as failed or disconnected, run /mcp again and re-authenticate.

Claude

Works on claude.ai and in the Claude desktop app. Custom connectors require a paid Claude plan (Pro, Max, Team, or Enterprise).

In Claude, open Settings → Connectors and click Add custom connector.
Paste the server URL, name it “agent-data,” and click Add:
```
https://agent-data.motie.dev/mcp
```
Click Connect, sign in with your agent-data account, and approve access.
Start a new chat and turn on the agent-data connector.

To stop Claude from asking before each tool call, set the connector’s Needs approval setting to Always allow.

ChatGPT

Enable Settings → Security and login → Developer mode. Availability depends on your account and workspace policy. These steps follow OpenAI’s MCP connection guide.

Open ChatGPT Plugins and select the + button.
Enter the connection details and create it:
- Name: agent-data
- Description: Discover and call public APIs, and request additional endpoints for the data you need.
- Connection: public MCP server at https://agent-data.motie.dev/mcp
Complete OAuth sign-in with your agent-data account, approve access, and review the discovered tools.
Start a new chat and add agent-data from its tools menu.

Cursor

Remote MCP servers with OAuth need Cursor 1.0 or later.

Open Cursor Settings → MCP and click New MCP Server. This opens ~/.cursor/mcp.json.

Add agent-data and save:

{
  "mcpServers": {
    "agent-data": {
      "url": "https://agent-data.motie.dev/mcp"
    }
  }
}

Back in the MCP settings, agent-data appears with a login prompt. Click it, sign in with your agent-data account, and approve access.

Troubleshooting: If the tools don’t show up, toggle the server off and on in the MCP settings, or restart Cursor.

VS Code

Requires VS Code 1.102 or later with GitHub Copilot.

Run this in a terminal:
Terminal window
```
code --add-mcp '{"name":"agent-data","type":"http","url":"https://agent-data.motie.dev/mcp"}'
```
Or add it from the editor: open the Command Palette, run MCP: Add Server, choose HTTP, and paste https://agent-data.motie.dev/mcp.
When VS Code asks to authenticate, allow it and complete the sign-in with your agent-data account in the browser.
Open Copilot Chat in agent mode — the agent-data tools are available under the tools picker.

To share the server with your team, commit it to your repo in .vscode/mcp.json:

{
  "servers": {
    "agent-data": {
      "type": "http",
      "url": "https://agent-data.motie.dev/mcp"
    }
  }
}

Windsurf

Open Windsurf Settings → Cascade → MCP servers and choose Add custom server. This opens ~/.codeium/windsurf/mcp_config.json.

Add agent-data and save:

{
  "mcpServers": {
    "agent-data": {
      "serverUrl": "https://agent-data.motie.dev/mcp"
    }
  }
}

Refresh the server list, then complete the sign-in with your agent-data account when prompted.

Codex CLI

Add agent-data as a remote MCP server:

codex mcp add agent-data --url https://agent-data.motie.dev/mcp

The tools are available the next time you start codex.

Prefer editing config directly? Add this to ~/.codex/config.toml:

[mcp_servers.agent-data]
url = "https://agent-data.motie.dev/mcp"

Troubleshooting: If the server shows as disconnected, run codex mcp list to check its status, then codex mcp login agent-data to re-authenticate. If an older Codex release reports missing required issuer, update Codex before retrying. OAuth sign-in was verified with Codex CLI 0.155.1.

Gemini CLI

Add agent-data as a remote MCP server:

gemini mcp add --transport http agent-data https://agent-data.motie.dev/mcp

Inside Gemini CLI, run /mcp auth agent-data and complete the sign-in that opens in your browser.
Run /mcp to confirm the agent-data tools are listed.

Available tools

The MCP server supports API discovery, calls and adding endpoints to an existing public API:

Tool	Description
`search`	Search the API catalog with a descriptive query, such as “YouTube video details”.
`list`	Browse APIs available to your account.
`docs`	Inspect an endpoint’s routes, parameters, and response schemas.
`call`	Invoke a route and return live data.
`whoami`	Show the connected account identity and authentication details.
`extension_target`	Inspect an existing API and its current endpoint contract.
`extension_submit`	Request new endpoints using saved request and endpoint IDs.
`extension_status`	Follow your saved request through publication or a reported blocker.

See Add an endpoint for request fields and retry behavior. A submission acknowledgement does not mean the endpoint is published. New-source builds are available through the CLI parse command and the dashboard.