Setup
Install in one sentence.
Today you install by pasting one invite sentence into your AI assistant, such as Claude Code or Cursor. If you'd rather install the client yourself, hook up an editor, or call the API from code, pick that path below.
Who can see your memory? Read the data access page.
Pick a path
Each path has a short, step-by-step walkthrough: what to copy, where to paste, how to confirm it works, and what to try if something goes wrong.
Paste the install sentence into the chat. The bot signs itself up, saves its own key, and checks that it works. Paste-into-chat walkthrough
Install the client with one Terminal command, then add it to your PATH if the shell can't find it. Mac walkthrough
Install with PowerShell. You get a brain command and an uninstall. Windows walkthrough
Add the MCP server to your project so Claude Code can find, add, and delete notes. Claude Code walkthrough
Add the same MCP server in Cursor. Cursor walkthrough
Mac and Windows config files, with the full path to the server. Claude Desktop walkthrough
Add B.R.A.I.N. to Gemini CLI's settings file, then confirm it with /mcp list. Gemini CLI walkthrough
Call the HTTP API yourself with curl, JavaScript, or Python, or install the published packages. API walkthrough
Add the hosted connector so ChatGPT, Claude.ai, and their phone apps can find and save notes. Connector walkthrough
Which bots work with this?
Grok Bot, Claude Code, and Cursor work today with the paste-into-chat path. ChatGPT, Claude.ai, and their phone apps can use the hosted connector. A chatbot that can only talk and can't run commands cannot use the paste-into-chat path.
You can still install it yourself on Mac or Windows, call the API directly, or use the hosted connector if the app supports custom connectors.
The public bot playbook is at /INSTALL-FOR-BOTS.md. A short index for automated clients is at /llms.txt. These are ordinary public pages, not hidden instructions.
1. Get your invite sentence and install your bot
This is the paste-into-chat path for AI assistants, including Grok Bot, Claude Code and Cursor. For the full walkthrough, including how to confirm it in a new chat and common fixes, see the AI bots guide.
Sign in
Open the dashboard and copy the team install sentence. You can do this before you subscribe. Subscribe in the dashboard when you are ready. Billed monthly. Cancel anytime.
Paste it into the chat
Open your AI assistant (for example Claude Code or Cursor), paste the install sentence as a new message, and send it. No extra prompt needed. Do not add the API key. The bot saves its own key in a private place and runs the installer.
Wait for the all-clear
The bot should reply
B.R.A.I.N. installed. Then press Refresh on the dashboard. The bot should show as Connected.
Install B.R.A.I.N. from https://brain.telonyte.com/i/<your-invite>
You: Install B.R.A.I.N. from https://brain.telonyte.com/i/<your-invite> Your bot: B.R.A.I.N. installed
2. What the bot does
Redeem
The bot opens the invite link and posts its stable bot id to get its own key.
Store
It saves the URL and key in its own secret store. The key is never printed in chat.
Check
It runs the installer, then
brain selftest, and repliesB.R.A.I.N. installed.
Re-running with the same bot id reuses its key and seat. A new bot id uses another bot from your plan. After it replies, press Refresh on the dashboard. The bot should show as Connected.
How to use what comes back
When a bot looks something up, it should answer only from what B.R.A.I.N. returns. If a fact is missing or only partly there, the bot should say what is there and that the rest is not in memory. It should not guess or fill gaps from its own knowledge. All bots on the team share the same memory.
AI agent memory
How this memory is different
B.R.A.I.N. is memory your bots use, not a model that answers for them. A lookup returns only what is already stored.
Memory uses local rules and search. We never call a model to write a note, extract a fact, or fill a gap.
If a fact is missing or only partly there, that is all your bot gets back. Do not fill the gap from the bot's own knowledge.
Your bot gets back only the facts that matter from what is already in memory, so its own model reads less.
Several bots on one team share one brain. Each bot has its own key, which you can pause or revoke. Other teams cannot see yours.
For developers
You don't have to go through a bot. You can call B.R.A.I.N. yourself from a terminal, a script, or an editor. Each path has a step-by-step walkthrough, and the Developers page keeps the short copy-paste reference:
- Mac walkthrough · short reference
- Windows walkthrough · short reference
- Claude Code, Cursor, Claude Desktop, and Gemini CLI walkthroughs
- Direct API walkthrough, with examples in curl, JavaScript, and Python
- ChatGPT and Claude.ai connector
The installer works on Mac and Linux (install.sh) and on Windows (install.ps1). Both need Python 3.
Install the JavaScript and Python packages
The JavaScript and Python packages are published. Install them with the commands below, or call the HTTP API yourself with no package at all. The MCP server is available now with the client. LangChain and LlamaIndex adapters are published too. Package pages: @telonyte/brain, telonyte-brain, @telonyte/brain-langchain, @telonyte/brain-llamaindex, telonyte-brain-langchain, and telonyte-brain-llamaindex.
npm install @telonyte/brain
pip install telonyte-brain
npm install @telonyte/brain-langchain pip install telonyte-brain-langchain
npm install @telonyte/brain-llamaindex pip install telonyte-brain-llamaindex
# installed with the client, not an npm package
brain-mcp --help
The MCP server comes with the installer. See editor setup for the config to paste.
Keep keys private
Never paste an API key into chat, a ticket or a note. Each bot has its own key, and you can pause or revoke it in the dashboard. Questions? Contact us.
Ready to install?
Get your team install sentence from the dashboard.