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

# MCP Connect

> Bring Corvic into Claude, ChatGPT, Grok, Muse, your IDE, or your own code over the Model Context Protocol.

**MCP Connect** puts Corvic inside the AI tools you already use. Add Corvic as an MCP server, and your assistant can list your rooms, ask your memories, run pipelines and playbooks, read documents through lenses, and pull results back as tables or files, all with your Corvic permissions.

The [Model Context Protocol](https://modelcontextprotocol.io) (MCP) is an open standard for connecting AI assistants to tools and data. Corvic serves one MCP endpoint for every organization:

```text theme={null}
https://api.corvic.ai/mcp
```

[Same Claude, More Context: Connecting Claude to Your Enterprise Data](https://www.youtube.com/embed/iKiAdKNUTjE)

## Two ways to connect

| | Sign in with Corvic | Access token |
| - | - | - |
| **How** | Paste the URL; your assistant opens Corvic's sign-in page and you approve | Send `Authorization: Bearer <token>` with every request |
| **Best for** | Claude, ChatGPT, Grok, Muse, Cursor, VS Code: anything with a "custom connector" screen | Scripts, backends, agent frameworks, and clients that only take headers |
| **What it reaches** | Everything your account can reach, in the organization you sign in to | What the token was created for: your whole organization, or one room |
| **Set up** | Nothing to create | [Create an access token](https://docs.corvic.ai/mcp-connect/access-tokens.md) |

The organization a call reaches comes from your sign-in or token, never from the URL, so the same address works for everyone.

## Connect your assistant

> [card] [Claude](https://docs.corvic.ai/mcp-connect/clients/claude.md)
>
> Claude.ai, Claude Desktop, and Claude Code.

> [card] [ChatGPT](https://docs.corvic.ai/mcp-connect/clients/chatgpt.md)
>
> A custom connector in developer mode.

> [card] [Grok](https://docs.corvic.ai/mcp-connect/clients/grok.md)
>
> Grok connectors, or remote MCP tools in the xAI API.

> [card] [Meta Muse](https://docs.corvic.ai/mcp-connect/clients/muse.md)
>
> A custom connector in Muse.

> [card] [IDEs and coding agents](https://docs.corvic.ai/mcp-connect/clients/ides.md)
>
> Cursor, VS Code, Claude Code, and Codex.

> [card] [Your own code](https://docs.corvic.ai/mcp-connect/clients/code.md)
>
> Python, TypeScript, and agent frameworks.

## What your assistant can do

| Area | For example |
| - | - |
| **Rooms and data** | List rooms, create a data source, upload files into a table |
| **Pipelines** | Build a pipeline, run it, schedule it, and watch its status |
| **Memories** | Ask a [Virtual Memory](https://docs.corvic.ai/features/virtual-memory.md) in words, or search its knowledge graph precisely |
| **Document lenses** | Read documents into fields through a [Document Lens](https://docs.corvic.ai/features/document-lens.md) |
| **Agents and playbooks** | Hand open-ended work to the room's agent, or run a [playbook](https://docs.corvic.ai/features/playbooks.md) |
| **Results** | Fetch rows, files, and short-lived share links |

See the full list in [Room tools](https://docs.corvic.ai/mcp-connect/room-tools.md). To talk to one deployed agent instead of a whole room, use [Agent MCP](https://docs.corvic.ai/mcp-connect/agent-mcp.md).

> [!TIP]
>
> Try a tool before you connect anything: every room has an [MCP Playground](https://docs.corvic.ai/mcp-connect/playground.md) that lists the tools and runs them live.

## The other direction

MCP Connect brings Corvic **into** other tools. To let Corvic's agents call **other** MCP servers, add them in the [Connector Vault](https://docs.corvic.ai/connector-vault/mcp-servers.md).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.