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

# AI Assistants and API Access in LanderLab: Overview

> Connect LanderLab to Claude, ChatGPT, Cursor, and other AI assistants through MCP, or automate landing pages, leads, and analytics with the REST API.

LanderLab can be operated outside its own interface in two ways. **MCP** lets an AI assistant do the work in plain English. The **REST API** lets your own systems do it in code. Both use the same API key.

## Which one you need

Use **MCP** when a person is asking for something conversationally: pull last week's conversion numbers, publish a page, shift A/B traffic weights. Use the **REST API** when a system needs to do it repeatedly and reliably without anyone watching.

<Columns cols={2}>
  <Column>
    <Card title="MCP for AI assistants" icon="comments" href="/mcp/overview">
      Manage landing pages, variants, leads, and analytics in plain English from any MCP-compatible assistant.
    </Card>
  </Column>

  <Column>
    <Card title="REST API" icon="code" href="/api-overview">
      Programmatic access to landers, variants, leads, integrations, folders, domains, and analytics.
    </Card>
  </Column>
</Columns>

## Getting an API key

Both paths start here. Keys are scoped to your organization and passed in the `X-API-Key` header.

<Card title="Generate a LanderLab API key" icon="key" horizontal href="/mcp/generate-api-key">
  Create a key, find your existing keys, and keep them secure.
</Card>

<Warning>
  Treat an API key like a password. Anyone holding it can read and modify your landing pages and leads. Never paste one into a shared document, a screenshot, or a support ticket.
</Warning>

## Connecting an AI assistant

Setup takes a few minutes and differs slightly per client. Pick yours:

<Columns cols={2}>
  <Column>
    <Card title="Claude" icon="message" href="/mcp/ai-assistants/claude">
      [Claude.ai](/mcp/ai-assistants/claude), [Claude Desktop](/mcp/ai-assistants/claude-desktop), and [Claude Code](/mcp/ai-assistants/claude-code).
    </Card>
  </Column>

  <Column>
    <Card title="ChatGPT" icon="comment" href="/mcp/ai-assistants/chatgpt">
      Add LanderLab as a connector and manage pages from a chat.
    </Card>
  </Column>
</Columns>

<Columns cols={2}>
  <Column>
    <Card title="Coding assistants" icon="terminal" href="/mcp/ai-assistants/cursor">
      [Cursor](/mcp/ai-assistants/cursor), [Windsurf](/mcp/ai-assistants/windsurf), and [Gemini CLI](/mcp/ai-assistants/gemini-cli).
    </Card>
  </Column>

  <Column>
    <Card title="Manus" icon="robot" href="/mcp/ai-assistants/manus">
      Add your key through the settings UI or a JSON config.
    </Card>
  </Column>
</Columns>

Any MCP-compatible client works, not only the ones documented here. If yours is not listed, the [MCP overview](/mcp/overview) has the server URL and authentication details you need.

## What you can do through MCP

Once connected, you can ask for things directly instead of clicking through the dashboard:

* Read workspaces, landing pages, variants, leads, and analytics
* Publish pages and edit variants
* Adjust A/B test traffic weights
* Update leads and configure integrations

This is most useful for reporting and quick operational changes. Asking for last month's conversion rate by device across five pages is one sentence, versus several minutes of navigation.

<Note>
  Tool permissions are set in your assistant, not in LanderLab. Review what you allow before granting write access, since mutation tools can publish and modify live pages.
</Note>

## What you can do through the API

The REST API covers the same surface with finer control. Common uses:

* Create landing pages programmatically, including [importing from a URL](/api-reference/landers/create-lander-from-url)
* [Publish](/api-reference/landers/publish-lander) and unpublish pages from your own deployment process
* Pull [leads](/api-reference/leads/list-organization-leads) into an internal database or data warehouse
* Query [unified analytics](/api-reference/analytics/get-unified-analytics) across workspaces, pages, and variants
* Manage [A/B variants and weights](/api-reference/ab-testing/set-variant-weights-ab-traffic-split) automatically based on performance

<Card title="API overview" icon="book" horizontal href="/api-overview">
  How authentication works, what resources exist, and how the endpoints are organized.
</Card>

<Tip>
  Call [get current context](/api-reference/account/get-current-context) first. It returns the organization and workspace IDs that almost every other endpoint requires.
</Tip>