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

# One brain in every AI tool

A chat history stays inside the tool that holds it. Connect those tools to one brain and each of them
can search the same pages, read the same sources and add to them. You can capture a note from a
terminal agent and ask about it later in ChatGPT.

## See it in each tool

These example sessions come from [livingbrain.com](https://livingbrain.com).

#### Claude

![Claude calling living\_brain\_show\_graph and showing the brain's 3D graph inside the chat](/_fern-files/_img/2d3fb53de3fdd6f6144663c82d853a5906126f889f590c8e5f07028a55f7c97d.webp)

**Open the graph inside the chat.** `living_brain_show_graph` draws the brain's pages and the
connections between them. You can move around it without leaving the conversation.

Claude connects with your account, so there's no API key to manage. See
[MCP with sign-in](/mcp/sign-in).

#### ChatGPT

![ChatGPT listing brains, indexing pages and reading a page before answering](/_fern-files/_img/67b6e64d9656bd3dd3d7f411eb98a3562b028bd4f93248fb4a1d835d14ca16aa.webp)

**Watch it check before it answers.** ChatGPT lists the available brains, reads the index, opens
the page and then answers with that page's title and last update.

ChatGPT also connects with your account. See [MCP with sign-in](/mcp/sign-in).

#### Cursor

![Cursor CLI recalling a decision from Living Brain and answering from memory](/_fern-files/_img/95f19831810a6e899ad727e1697ee388d87f29c7804c0d26b52bf53861a41ccd.webp)

**Ask what the team decided.** The agent finds the decision, says where and when it was made, and
drafts its answer from that page.

Cursor connects with an API key. See [MCP with API key](/mcp), or the
[coding agent example](/docs/use-cases/coding-agent-memory).

#### Gemini CLI

![Gemini CLI answering the same onboarding question from Living Brain](/_fern-files/_img/fd171a2ea301bf7b7629df28d0c412257f14ac82cac4b5d6720b5d4a513b0ce8.webp)

**Ask the same question and get the same page.** Gemini CLI reads the same decision as Cursor,
because both tools are looking at one brain.

Terminal agents can use the [Living Brain CLI](/cli).

#### Hermes Agent

![Hermes Agent running the Living Brain CLI to find a brain and capture a note](/_fern-files/_img/55ece85b8d2042abdc21cc53256240f66d246587635f02aae29a3eb55227dba9.webp)

**Save notes from an agent.** Hermes looks up the brain by name with the CLI and captures the note.
The capture returns as `pending` while the brain compiles it in the background.

See the [CLI guide](/cli).

## Connect your tools

| Tool                                               | Connects through                 | Sign in with                |
| -------------------------------------------------- | -------------------------------- | --------------------------- |
| Claude, ChatGPT                                    | [MCP with sign-in](/mcp/sign-in) | Your Living Brain account   |
| Claude Code, Codex, Cursor                         | [MCP with API key](/mcp)         | An API key and a subject ID |
| Gemini CLI, Hermes Agent and other terminal agents | [Living Brain CLI](/cli)         | An API key and a subject ID |

Use the same brain and the same subject in each tool. A different subject means a different set of
brains.

## Prompts to try

Once a tool is connected, ask for the brain by name.

```text
Save these notes to my Research brain.

What's the latest note in my Research brain?

What did we decide about the onboarding flow? Show me the page it came from.

Which pages connect to the onboarding flow page?

Show me the graph of my Research brain.
```

The last prompt opens the interactive view on hosts that support it. Other tools can return the same
pages and connections as text.

## What the tools can do

Sign-in hosts get these tools. The [API-key server](/mcp#tools) has one tool for each API operation.

| You want to                             | Tool                            |
| --------------------------------------- | ------------------------------- |
| See your brains                         | `living_brain_list_brains`      |
| Find pages by meaning or keyword        | `living_brain_search`           |
| Read a page with its sources            | `living_brain_read_page`        |
| Skim a brain's pages, pinned ones first | `living_brain_get_index`        |
| See how pages relate                    | `living_brain_list_connections` |
| Open the 3D graph                       | `living_brain_show_graph`       |
| Save a note, file, link or transcript   | `living_brain_capture`          |
| Check whether a capture has compiled    | `living_brain_capture_status`   |

## Limits to plan for

* **New captures take a moment.** A capture compiles in the background, so a search straight afterwards
  won't find it. Use `living_brain_capture_status` to check.
* **Everyone with access reads the same pages.** Access is per brain, so keep private material in its
  own brain.
* **Tools decide when to call the brain.** Name the brain in your prompt, or add an instruction to the
  tool's custom instructions or rules file.

[Coding agent memory](/docs/use-cases/coding-agent-memory) shows the same setup for a shared code project.