Model Context Protocol

Connect Veille to Claude Code, Cursor, Gemini CLI, OpenAI Developer Mode, and other MCP clients using your API key.

The Model Context Protocol (MCP) lets AI assistants call Veille as tools: same operations as the REST API, with your API key. You can document setup like many teams do: a short README, a sample mcp.json, and copy-paste steps for your users.

What you need

A Veille account (register if needed) and access to the dashboard.
An API key with the x-api-key header (see API Keys).
MCP endpoint: https://mcp.veille.io/

How it works

The server uses Streamable HTTP at https://mcp.veille.io/. It is not a normal web page. Each tool maps to an API operation; billing and behaviour match the REST API. Use the rest of this documentation for endpoints and parameters.

Claude Code

Official docs: Model Context Protocol (MCP). Example: claude mcp add --transport http veille https://mcp.veille.io/ with --header x-api-key: YOUR_API_KEY. Use the same URL as in the Cursor section below.

Cursor

See the Cursor MCP documentation. In Settings → MCP, add an HTTP server, or use a project mcp.json with a url (ending in /) and headers for your API key.

Use the MCP endpoint for Veille below. Never commit secrets. Prefer environment-specific config or a placeholder like YOUR_API_KEY.

{
  "mcpServers": {
    "veille": {
      "url": "https://mcp.veille.io/",
      "headers": {
        "x-api-key": "YOUR_API_KEY"
      }
    }
  }
}

Gemini CLI

See MCP server in the Gemini CLI documentation. Point the client at https://mcp.veille.io/ and pass your API key (for example via x-api-key) as the tool docs describe.

OpenAI Developer Mode

See Developer mode in the OpenAI API documentation. Wire MCP tools to https://mcp.veille.io/ and your API key following that guide.