Algolia MCP Server

algoliasearchanalyticsrecommendoauthread-only

Summary

Algolia's managed MCP server at mcp.algolia.com — read-only access to indices, Recommend models and search analytics over OAuth, with no API keys to manage.

Features

List indices, run searches and do facet-value search across your application
Retrieve Recommend model output — related products and complementary items
Search analytics: top searches, no-result queries, no-click rates, click positions, filter usage
User and country metrics for query traffic
OAuth login with no client ID or secret to manage; inherits your Algolia account permissions
Strictly read-only — cannot create indices or modify and delete records

Installation

Set up this MCP server in your favorite AI agent — copy a ready-made configuration below.

Any MCP-compatible agent

Most agents (Claude, Cursor, Windsurf, VS Code, and more) read a standard mcpServers configuration.

Open your agent's MCP configuration file.
Merge the snippet below into it.
Restart the agent — the "Algolia MCP Server" tools will be available.

{
  "mcpServers": {
    "algolia-mcp-server": {
      "command": "claude",
      "args": [
        "mcp",
        "add",
        "--transport",
        "http",
        "algolia",
        "https://mcp.algolia.com/mcp"
      ]
    }
  }
}

Claude Code

Add the server with a single CLI command.

claude mcp add algolia-mcp-server -- claude mcp add --transport http algolia https://mcp.algolia.com/mcp

Claude Desktop

Add the server to claude_desktop_config.json (Settings → Developer → Edit Config).

{
  "mcpServers": {
    "algolia-mcp-server": {
      "command": "claude",
      "args": [
        "mcp",
        "add",
        "--transport",
        "http",
        "algolia",
        "https://mcp.algolia.com/mcp"
      ]
    }
  }
}

Cursor

Cursor also accepts a plain URL entry in .cursor/mcp.json: {"mcpServers":{"algolia":{"url":"https://mcp.algolia.com/mcp"}}}

{
  "command": "npx",
  "args": [
    "-y",
    "mcp-remote",
    "https://mcp.algolia.com/mcp"
  ]
}

Description

Algolia runs its MCP server as a managed service at https://mcp.algolia.com/mcp, so there is nothing to install and no key to rotate. Authentication is OAuth against your Algolia account, and — unusually — you do not supply an OAuth client ID or secret; the server inherits whatever application and index permissions your user already holds.

The tool surface covers three things a search team actually asks about. Search operations: list the indices in an application, run queries against one, and do facet-value search. Recommendations: pull output from Recommend models, such as related products and complementary items. Analytics: top searches, queries returning no results, no-click rates, click positions, filter usage, and user and country breakdowns — the numbers you would otherwise pull from the dashboard one panel at a time.

The important constraint is that it is read-only by design. The server cannot create indices, modify records or delete anything. That makes it safe to hand to an analyst or to leave connected in a shared workspace, and it makes it the wrong tool for indexing pipelines — those still go through the standard API clients.

Algolia publishes two variants: the Productivity MCP described here, for user-scoped exploration and analysis, and a Public MCP scoped to selected indices and Recommend models within a single application. MCP must remain enabled in the Algolia dashboard for either to connect.

Setup takes one command in Claude Code, a URL entry in .cursor/mcp.json for Cursor, or a custom connector on claude.ai for the web client.

Best for teams debugging relevance — "which queries return nothing", "what are people searching that we do not stock" — where the questions are exploratory and the answers live across search, analytics and Recommend at once.