> ## Documentation Index
> Fetch the complete documentation index at: https://www.algolia.com/llms.txt
> Use this file to discover all available pages before exploring further.

# MCP Server (Model Context Protocol)

> Connect AI assistants to Algolia indices using the Algolia MCP Server.

The Model Context Protocol (MCP) is an open standard that lets AI agents call external tools
and retrieve structured data in real time.
Algolia supports MCP with two offerings.
Teams can connect assistants to their data without building new integrations for each model or workflow.

**Both are Algolia-managed services.**

* [Algolia Public MCP](#algolia-public-mcp). **Application-scoped search and recommendations** for selected indices and Recommend models in one Algolia application. Use it to expose data to external or customer-facing AI assistants.
* [Algolia Productivity MCP](#algolia-productivity-mcp). **User-scoped exploration and analysis** for people in your organization across the Algolia applications and indices allowed by each user's permissions.

<Note>
  Both MCP servers are **read-only**: they search, recommend, and (for Productivity MCP) report analytics.
  Neither one creates or modifies indices, records, or settings.
  To make changes, use the [Algolia CLI](/doc/tools/cli/get-started), the [dashboard](https://dashboard.algolia.com), or an [API client](/doc/libraries/sdk).
</Note>

Use the following table to choose the MCP server that fits your use case:

| Need                                                                                      | Use                      |
| ----------------------------------------------------------------------------------------- | ------------------------ |
| Expose selected indices and Recommend models to external or customer-facing AI assistants | Algolia Public MCP       |
| High-traffic, low-latency, customer-facing agents                                         | Algolia Public MCP       |
| Let people in your organization analyze data allowed by their Algolia permissions         | Algolia Productivity MCP |
| LLM-guided analysis of searches, analytics, and index data                                | Algolia Productivity MCP |

## Algolia Public MCP

What it's for:

* **AI-facing search and recommendations** on a defined set of indices.
* Integrate with MCP-compatible platforms like [ChatGPT](https://chatgpt.com), [Claude](https://claude.ai), [Vertex AI](https://cloud.google.com/vertex-ai), [Perplexity](https://www.perplexity.ai), [Amazon Q](https://aws.amazon.com/q/),
  [Microsoft Copilot](https://www.microsoft.com/en-us/microsoft-copilot), or [Salesforce Agentforce](https://www.salesforce.com/agentforce/).
* Support production use cases such as product discovery, support bots, and customer-facing AI experiences.

Key traits:

* **Tool-scoped interface** that exposes search and recommend without exploding tool counts.
* Usage counts toward your existing Algolia plan. There's no separate MCP fee.

To learn more, see [Get started with Algolia Public MCP](/doc/guides/model-context-protocol/public-mcp).

## Algolia Productivity MCP

What it's for:

* **Exploration, analysis, and internal productivity** with LLM-powered tools.
* Personal workflows where your own Algolia permissions determine access.
* Business intelligence-style questions and one-off operational checks like spotting trending searches or zero-result queries.

Key traits:

* **Enable once per account.** Productivity MCP is enabled once for the account. This turns the feature on.
* **Sign in per user.** Signing in is separate from enabling. Each user signs in with their own Algolia login. This identifies them and sets what they can access.
* **Read-only index access**, plus analytics tools.
* **Registry-visible.** It appears in public MCP directories.

To learn more, see [Get started with Algolia Productivity MCP](/doc/guides/model-context-protocol/productivity-mcp).