Skip to main content
Emerge provides two MCP (Model Context Protocol) servers for different purposes. Connect your AI tools to search documentation, query user data, or both.

​
Two MCP Servers

Docs MCP Server

Search and query this documentation
  • Semantic documentation search
  • API reference lookup
  • Code example retrieval
  • Best for: Learning the API, troubleshooting

Query MCP Server

Query consented user data via AI tools
  • Search history, browsing, YouTube, ad interactions, receipts
  • Session-based credentials
  • Direct data access for AI agents
  • Best for: Building AI applications with user data
ServerPurposeURLAuth Required
Docs MCPSearch documentationhttps://docs.emergedata.ai/mcpNo
Query MCPQuery user datahttps://mcp.emergedata.ai/mcpYes (API token + uid)
Both use the same Model Context Protocol and work with Claude, Cursor, VS Code, and Claude Code.

​
Quick setup

Add documentation search to your AI tool:
# Claude Code
claude mcp add --transport http emerge https://docs.emergedata.ai/mcp

// Cursor (.cursor/mcp.json)
{
  "mcpServers": {
    "Emerge": {
      "type": "http",
      "url": "https://docs.emergedata.ai/mcp"
    }
  }
}
See Docs MCP Setup for all tools.

​
What you can do

Ask questions

“How do I query user search history?” — AI searches docs and returns the answer with code examples

Generate code

“Generate Python code to create a signed link” — AI finds SDK page and provides working implementation

Query user data

“Get search history for the last week” — AI calls the Query MCP tool and returns results

Debug issues

“Why is my webhook signature invalid?” — AI finds troubleshooting info and suggests fixes

​
How it works

Web SearchMCP Connection
May find outdated contentAlways current documentation
Results from multiple sourcesSingle source of truth
Generic code examplesEmerge-specific implementations
Manual copy/pasteStructured data for AI

​
One-shot prompt for app builders

Use this prompt in tools like Lovable, Cursor, or Claude Code after connecting MCP: 
You are building a full-stack app that integrates Emerge Link + Query.
Requirements:
- Use https://link.emergedata.ai/link/start to start consent.
- Generate HMAC-SHA256 signature over sorted raw params (exclude signature).
- Use signing_secret for HMAC; never expose secrets in the frontend.
- timestamp must be ISO 8601 (new Date().toISOString()).
- Generate random state, store it server-side, verify on callback.
- redirect_uri must be pre-approved in the Control Room.
- Never ask end-users for uid. If you send uid, use your internal user id. Otherwise store the callback uid.
- Query sync endpoints with uid: https://query.emergedata.ai/v1/sync/get_*
- Poll `GET /export/status/{uid}` and query only when the provider in `sources[]` is `data_ready: true`.
Return TypeScript and Python examples with async/await and error handling.

​
Contextual menu

Every page in these docs includes a contextual menu with one-click AI integrations. Look for the menu icon on any page to:
  • Copy as Markdown for pasting into AI chats
  • Open directly in Claude or ChatGPT
  • Copy the MCP server URL
Learn more in Contextual Menu.