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

# MCP overview

> Connect AI assistants to your Spotlight organization via the Model Context Protocol.

## What is Spotlight MCP?

Spotlight exposes an [MCP](https://modelcontextprotocol.io) (Model Context Protocol) server so AI assistants — such as Cursor, Claude Desktop, or other MCP-compatible clients — can read and manage your organization's Spotlight data using natural language.

MCP wraps the same **organization API** you use for HTTP integrations. Your API key, rate limits, and permissions are identical whether you call REST endpoints directly or invoke MCP tools.

<CardGroup cols={2}>
  <Card title="Connect a client" icon="link" href="/spotlight-mcp/connect">
    Set up Cursor, Claude Desktop, or another MCP client.
  </Card>

  <Card title="Tools reference" icon="wrench" href="/spotlight-mcp/tools">
    Full list of MCP tools and what each one does.
  </Card>

  <Card title="Best practices" icon="lightbulb" href="/spotlight-mcp/best-practices">
    How agents should work with prompts, quotas, and scheduling.
  </Card>

  <Card title="REST API" icon="code" href="/api-reference">
    Prefer HTTP? Start with the REST API introduction.
  </Card>
</CardGroup>

## Prerequisites

1. **API access enabled** — MCP is available when API access is enabled for your organization. In the Spotlight app, go to **Settings → API** and turn on the API.
2. **API key** — Copy your key from **Settings → API**. MCP uses the same key as REST calls.
3. **MCP endpoint** — Find your MCP URL under **Settings → MCP**. The default is:

```text theme={null}
https://app.get-spotlight.com/api/mcp
```

<Note>
  MCP is currently in closed beta. If you do not see the API or MCP tabs in Settings, contact your Spotlight account team to request access.
</Note>

## What you can do with MCP

| Area                  | Examples                                                                            |
| --------------------- | ----------------------------------------------------------------------------------- |
| **Discovery**         | List brands, fetch organization and brand context, read the built-in usage guide    |
| **Analysis data**     | Pull run history, per-prompt results, run stats, sources, and content suggestions   |
| **Perception**        | Read perception prompts and responses; add new perception prompts                   |
| **Prompt management** | Create topics and prompts, batch-import prompt sets, delete prompts or whole topics |

Typical workflows include asking an assistant to "summarize our latest run," "add five consideration-stage prompts under a new topic," or "export perception scores from the last report."

## How MCP relates to the REST API

Each MCP tool maps to one or more REST endpoints under `https://app.get-spotlight.com/api`. For example, `list_brands` calls `GET /v1/brands`, and `get_run_stats` calls `GET /v1/runs/{runId}/stats`.

Agents should call **`get_organization_context`** once per session and **`get_brand_context`** before working with a specific brand. See [Best practices](/spotlight-mcp/best-practices) for the recommended workflow.

## Transport and protocol

* **Transport:** HTTP (streamable HTTP MCP)
* **Authentication:** `x-spotlight-api-key` header (same as the REST API)
* **Scope:** One API key = one organization; all tools are scoped to that organization

## Next steps

1. [Connect your MCP client](/spotlight-mcp/connect)
2. [Review available tools](/spotlight-mcp/tools)
3. [Follow agent best practices](/spotlight-mcp/best-practices)