Developers

Use it from your own code.

Pasting the install sentence to a bot is the quick way in. If you'd rather call B.R.A.I.N. yourself, from a terminal, a script, or an editor, everything you need is on this page. Examples use placeholders like $BRAIN_KEY. Swap in your own.

Step-by-step walkthroughs

If you want numbered steps, a quick test, and common fixes, start here. This page stays the short copy-paste reference.

AI bots, including Grok Bot

Paste-into-chat walkthrough

Direct API

API walkthrough in curl, JavaScript, and Python

Get a key

Sign in, copy your team's install sentence from the dashboard, and have a bot redeem it. Or open the invite link yourself and send it a bot name. Either way you get a key for that bot, and it's shown once. Save it as BRAIN_KEY in your own secret store. Never paste it into a chat.

Install the client

The client is a small Python program. It needs Python 3, and nothing else. The installer saves your address and key to a private config file, so later commands work without any environment variables. Prefer a full walkthrough? See the Mac or Windows guide.

Mac and LinuxAvailable now
# set BRAIN_KEY first, then:
curl -fsSL https://brain.telonyte.com/install.sh | bash
brain selftest
Windows (PowerShell)Available now
# set $env:BRAIN_KEY first, then:
irm https://brain.telonyte.com/install.ps1 | iex
brain selftest

If the shell says brain was not found

On Mac and Linux the client goes in ~/.local/bin. Run it by its full path, then add that folder to your PATH.

Mac and Linux
~/.local/bin/brain selftest

# zsh
echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.zshrc
# bash (use ~/.bash_profile on a Mac)
echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.bashrc

You can also run the installer with BRAIN_ADD_PATH=1 and it adds that line to your profile for you. To install somewhere else, set BRAIN_INSTALL_DIR.

Windows (PowerShell)
# if PATH was not updated, run it by its full path:
python $env:LOCALAPPDATA\brain-hosted\brain.py selftest

# to add the folder to your user PATH, install with:
$env:BRAIN_ADD_PATH = "1"

Uninstall

Uninstalling removes the client and any PATH line the installer added. Your saved config stays unless you ask for it to go too, so your key isn't deleted by surprise.

Mac and Linux
curl -fsSL https://brain.telonyte.com/install.sh | bash -s -- --uninstall
Windows (PowerShell)
$env:BRAIN_UNINSTALL = "1"
irm https://brain.telonyte.com/install.ps1 | iex

# or, with the switch:
& ([scriptblock]::Create((irm https://brain.telonyte.com/install.ps1))) -Uninstall

To remove the saved config as well, set BRAIN_REMOVE_CONFIG=1 before you run the uninstall.

Speed and recall

Hosted

Round trip over the internet from a computer in the eastern US. About 50 ms for a typical recall. 95 of every 100 come back in under 125 ms.

Right answer

On LongMemEval, 118 questions about months of chat, the part that held the answer was in the result 91.5% of the time (108 of 118) with the local search model on, and 82.2% (97 of 118) with the hosted engine only. On our 70-question test it was 58 of 70 (50 with the hosted engine only). On the full 776-question set it was 70.2%.

On your computer

Each bot's computer keeps its own encrypted copy of your team memory, so a lookup is answered right there. 1.2 to 4.6 ms for the lookup itself, and about 30 ms for the whole brain find command.

Hive sync

A note saved by one bot reaches the copy on every other computer on your team in about a third of a second. 95 of every 100 arrive in under 0.4 s.

The 1.2 to 4.6 ms lookup and 30 ms command times are for the command-line tool, from our own staging runs on November 14, 2026 (the lookup itself, then the whole brain find command). In the JavaScript and Python packages on production, a local lookup (the lookup itself, not the whole command) was 0.15 ms and 0.2 ms, compared with about 50 ms from the host. On LongMemEval, the part that held the answer was in the result 91.5% of the time (108 of 118) with the local search model on, and in the top 5 results 91.53% of the time, and in the top 20 96.61%. On the full LOCOMO set of 776 questions it was in the result 70.2% of the time (in the top 5 results 70.23%, and in the top 20 87.63%). On big team memories (7,164 to 70,847 notes), a typical lookup measured in-process on our own machine on November 16, 2026 took 3.312 to 21.394 ms, down from 35.319 to 412.048 ms before, and the right note came back at least as often. Times vary with your computer and network. These recall figures measure whether the part of the chat that held the answer came back, not model-graded answers. Watch a real timed recall in the speed demo on the home page.

Command line

Commands
brain find staging deploy
brain add my-bot "NEW | The staging deploy runs on Fridays | team notes"
brain delete <uid>
brain status
brain selftest
brain outbox --push

brain add prints a uid for the note it saved. Pass that to brain delete to remove the note. If the host can't be reached, brain add keeps the note in an outbox and brain outbox --push sends it later.

HTTP API

The API lives at the same address as this site. Send your key as a bearer token and ask for JSON. You can call it with curl or a few lines of JavaScript or Python, with no package required. If you'd rather use a package, the clients are published: npm install @telonyte/brain and pip install telonyte-brain. Package pages: npm and PyPI. LangChain and LlamaIndex adapters are published too: npm install @telonyte/brain-langchain, npm install @telonyte/brain-llamaindex, pip install telonyte-brain-langchain, and pip install telonyte-brain-llamaindex. Adapter pages: LangChain on npm, LlamaIndex on npm, LangChain on PyPI, and LlamaIndex on PyPI. The packages answer from a local copy on your computer, kept in sync with your team, and show Hyper speed on when they do. On serverless platforms they ask the host instead.

Headers on every call
Authorization: Bearer $BRAIN_KEY
content-type: application/json
GET /v1/status

Checks the connection. Returns { "ok": true, "tenant_id": "...", "notes": 12 }.

POST /v1/find

Body {"q":"staging deploy","n":5,"format":"json"}. Read the lines array in the reply.

POST /v1/add

Body {"bot":"my-bot","text":"NEW | the fact | source"}. The reply has the note's uid.

POST /v1/delete

Body {"uid":"..."}. Removes that note.

GET /v1/ws

Lists what your bots have saved. Add ?bot=my-bot to narrow it.

Passwords, codes, keys, account numbers and street addresses are refused when you add a note. Nothing is saved, so reword it and try again. Answer only from what find returns. If a fact is missing, say so.

The same example three ways

Prefer numbered steps? See the API walkthrough.

Each one saves a note, then looks it up.

curl
curl -s https://brain.telonyte.com/v1/add \
  -H "Authorization: Bearer $BRAIN_KEY" \
  -H "content-type: application/json" \
  -d '{"bot":"my-bot","text":"NEW | The staging deploy runs on Fridays | team notes"}'

curl -s https://brain.telonyte.com/v1/find \
  -H "Authorization: Bearer $BRAIN_KEY" \
  -H "content-type: application/json" \
  -d '{"q":"staging deploy","n":5,"format":"json"}'
JavaScript (fetch)
const base = "https://brain.telonyte.com";
const headers = {
  Authorization: `Bearer ${process.env.BRAIN_KEY}`,
  "content-type": "application/json",
};
const post = (path, body) =>
  fetch(base + path, { method: "POST", headers, body: JSON.stringify(body) }).then((r) => r.json());

await post("/v1/add", { bot: "my-bot", text: "NEW | The staging deploy runs on Fridays | team notes" });
const found = await post("/v1/find", { q: "staging deploy", n: 5, format: "json" });
console.log(found.lines.join("\n"));
Python 3 (standard library)
import json, os, urllib.request

BASE = "https://brain.telonyte.com"
HEADERS = {
    "Authorization": "Bearer " + os.environ["BRAIN_KEY"],
    "content-type": "application/json",
    "User-Agent": "brain-setup-example",
}

def post(path, body):
    req = urllib.request.Request(BASE + path, data=json.dumps(body).encode(), headers=HEADERS, method="POST")
    with urllib.request.urlopen(req) as r:
        return json.load(r)

post("/v1/add", {"bot": "my-bot", "text": "NEW | The staging deploy runs on Fridays | team notes"})
found = post("/v1/find", {"q": "staging deploy", "n": 5, "format": "json"})
print("\n".join(found["lines"]))

Keep the User-Agent header in the Python example. Some networks reject Python's default one.

MCP and editor setup

The installer also puts an MCP server on your machine, so an editor can use B.R.A.I.N. directly. It's available now. It isn't published to npm. Your editor starts it, and it gives the editor four tools: find, add, status, and delete. Step-by-step walkthroughs: Claude Code, Cursor, Claude Desktop, and Gemini CLI.

After you install, brain-mcp is on your PATH. If it isn't, use ~/.local/bin/brain-mcp on Mac and Linux, or %LOCALAPPDATA%\brain-hosted\brain-mcp.py on Windows. It reads your address and key from the config file the installer wrote, or from BRAIN_URL and BRAIN_KEY. You never put the key in a chat.

Claude Code.mcp.json or your user MCP config
{
  "mcpServers": {
    "brain": { "command": "brain-mcp" }
  }
}
Cursor.cursor/mcp.json or MCP settings
{
  "mcpServers": {
    "brain": { "command": "brain-mcp" }
  }
}
Claude Desktop on Mac~/Library/Application Support/Claude/claude_desktop_config.json
{
  "mcpServers": {
    "brain": {
      "command": "python3",
      "args": ["/Users/YOU/.local/bin/brain-mcp"]
    }
  }
}
Claude Desktop on Windows%APPDATA%\Claude\claude_desktop_config.json
{
  "mcpServers": {
    "brain": {
      "command": "python",
      "args": ["C:\\Users\\YOU\\AppData\\Local\\brain-hosted\\brain-mcp.py"]
    }
  }
}
Gemini CLI~/.gemini/settings.json or .gemini/settings.json
{
  "mcpServers": {
    "brain": { "command": "brain-mcp" }
  }
}

In Gemini CLI, restart, then run /mcp list to confirm it is connected. The Gemini CLI walkthrough has the full steps.

Editors that don't see your PATH can use the python3 form above in any of these files. If you'd rather not rely on the config file, add "env": { "BRAIN_URL": "https://brain.telonyte.com", "BRAIN_KEY": "YOUR_KEY" } next to command, and keep that file out of source control.

Keep keys private

Don't paste an API key into a chat, a ticket, or a note. Every bot has its own key, and if one leaks you can pause or revoke it from the dashboard without touching the others. Stuck on something? Contact us.