> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.ada.cx/mcp/introduction/getting-started/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.ada.cx/_mcp/server.

# Getting started

Ada's MCP Server works with any MCP-compatible AI assistant. Pick your client and follow the walk-through to connect.

## Pick your client

| Client                    | When to use it                                                                       | Walk-through                                                             |
| ------------------------- | ------------------------------------------------------------------------------------ | ------------------------------------------------------------------------ |
| Claude                    | Interactive analysis in Claude Desktop, or terminal-based automation in Claude Code. | [Connect Claude](/mcp/introduction/getting-started/claude)               |
| ChatGPT                   | Interactive analysis in the ChatGPT desktop app. Paid plan recommended.              | [Connect ChatGPT](/mcp/introduction/getting-started/chat-gpt)            |
| Codex CLI                 | Terminal-based analysis and automation in OpenAI Codex.                              | [Connect Codex CLI](/mcp/introduction/getting-started/codex-cli)         |
| Gemini CLI                | Terminal-based analysis for Gemini Enterprise users.                                 | [Connect Gemini CLI](/mcp/introduction/getting-started/gemini-cli)       |
| Google ADK                | Programmatic access from Python agents and backends.                                 | [Connect Google ADK](/mcp/introduction/getting-started/google-adk)       |
| Cursor / VS Code / custom | Any other MCP-compatible client.                                                     | [Other MCP clients](/mcp/introduction/getting-started/other-mcp-clients) |

Managing more than one AI Agent? See [Connecting multiple Agents](/mcp/introduction/getting-started/connecting-multiple-agents).

## Before you connect

* If you use OAuth, confirm your role. All dashboard roles can connect; Owner and Admin can call write tools, while Agent and Read Only are limited to read-only tools. See [Permissions](/mcp/introduction/authentication#permissions).
* Decide on [authentication](/mcp/introduction/authentication): OAuth (recommended for interactive clients) or API key (recommended for programmatic access). For Claude, add a [custom connector](/mcp/introduction/getting-started/claude).
* Review [data privacy](/mcp/introduction/data-privacy) settings for your assistant, especially model-training toggles.

## After you connect

* Try a prompt from the [prompt library](/mcp/prompt-library/overview).
* Explore available [tools](/mcp/tools/overview) to understand what the assistant can pull.
* Hit a problem? See [troubleshooting](/mcp/introduction/troubleshooting).

## Docs

- [Claude](https://docs.ada.cx/mcp/introduction/getting-started/claude.md)
- [ChatGPT](https://docs.ada.cx/mcp/introduction/getting-started/chat-gpt.md)
- [Codex CLI](https://docs.ada.cx/mcp/introduction/getting-started/codex-cli.md)
- [Gemini CLI](https://docs.ada.cx/mcp/introduction/getting-started/gemini-cli.md)
- [Google ADK](https://docs.ada.cx/mcp/introduction/getting-started/google-adk.md)
- [Other MCP clients](https://docs.ada.cx/mcp/introduction/getting-started/other-mcp-clients.md)
- [Connecting multiple Agents](https://docs.ada.cx/mcp/introduction/getting-started/connecting-multiple-agents.md)