Skip to content

TokaiApp/Tokai

BranchesTags

Repository files navigation

Tokai Logo

TOKAI

Neurosupportive Productivity · ADHD Management · Real-Time EEG Dashboard

神經支持型生產力 · ADHD 管理 · 即時 EEG 儀表板

Live Demo Website License Version AI

Try it live →

Overview 概述

Tokai is an open-source neurosupportive productivity application built for neurodivergent brains — especially people with ADHD. It provides a real-time neural dashboard that reads your cognitive state and helps you plan your day around how your brain is actually performing, not how you wish it were.

Tokai 是一款專為神經多樣性大腦（尤其是 ADHD 患者）打造的開源、神經支持型生產力應用程式。它提供即時神經數據儀表板，能讀取您的認知狀態，並協助您根據大腦的「實際」表現來規劃一天，而非盲目硬撐。

To the best of our knowledge, Tokai is the first app in the world to propose an AI task planner and agentic to-do list — TokAgent and TokTodo — driven by the user's own brain data.

據我們所知，Tokai 是全球首款提議根據使用者大腦數據來驅動 AI 任務規劃器與代理型待辦清單（TokAgent 與 TokTodo）的應用程式。

Neural data is simulated in this pre-alpha release. Integration with Muse 2 and other consumer EEG headsets is planned for future versions.

目前的 Pre-Alpha 版本中，神經數據為模擬狀態。計劃在未來版本中整合 Muse 2 及其他消費級 EEG 設備。

Features 核心功能

English 中文
🧠 Real-time neural dashboard — five live metric cards (Focus Index, Bio Energy, Neural Noise, T/B Ratio, Focus Window), each with an icon and status badge, updating every second 即時神經儀表板 — 五個即時指標卡片（專注指數、生理能量、神經噪訊、θ/β 比值、專注窗口），每秒更新，附圖示與狀態徽章
📊 Focus Stream — scrollable real-time line chart with ◀◀ / ◀ / ▶ / ▶▶ LIVE navigation buttons, an always-visible scrollbar, and reference lines for the 5-minute average, session average, and day average 專注串流 — 可捲動的即時折線圖，含 ◀◀ / ◀ / ▶ / ▶▶ LIVE 導覽按鈕、常駐捲軸，以及 5 分鐘均值、本次均值與當日均值參考線
🤖 TokAgent — AI assistant powered by Claude. Reads your live brain data, full task list, and journal entries to give context-aware task planning advice. Per-day sessions with timestamps. Bring your own Anthropic API key (stored locally, never on our servers) TokAgent — 由 Claude 驅動的 AI 助手，整合即時腦部數據、任務清單與日誌，提供情境感知的任務規劃建議。按日分組對話，附時間戳記。使用您自己的 Anthropic API 金鑰（僅本機儲存）
✅ TokTodo — full task manager with cognitive demand tags (Low / Med / High), emoji labels, time estimates, deadlines, creation timestamps, and a detail modal. Organized by day — past days are read-only TokTodo — 完整任務管理器，含認知負荷標籤（低／中／高）、表情符號標籤、預估時間、截止日期、建立時間戳記，以及任務詳情彈窗。按日分組，歷史日期唯讀
📓 TokNote — ADHD-friendly journal with multi-select mood tagging (10 moods: Hyperfocus, Flow, Focused, Restless, Scattered, Anxious, Fatigued, Zoned Out, Crashed, Low), Neural Insights embedded at the top, date + time stamps, and click-to-edit / delete for past entries TokNote — ADHD 友善日誌，支援多選情緒標籤（10 種），頂部嵌入神經洞察，附日期與時間戳記，可點擊既有條目進行編輯或刪除
📅 Day selector — dropdown menu to browse any past day. TokNote entries, TokAgent chat sessions, and TokTodo tasks are all filtered and displayed per selected day 日期選擇器 — 下拉選單瀏覽歷史日期，TokNote、TokAgent 對話與 TokTodo 任務均按所選日期篩選顯示
💊 TokMed — medication sidebar with a scrollable list (capped at 4 visible items) for tracking supplements and medications alongside your neural state TokMed — 藥物追蹤側欄，最多顯示 4 筆可捲動清單，與神經狀態並排呈現
🀄 Bilingual — full English and Traditional Chinese (繁體中文) support across all panels 雙語支援 — 所有面板完整支援英文與繁體中文
📱 Responsive — single-column stacked layout on mobile, three-panel row (TokNote · TokAgent · TokTodo) on desktop 響應式設計 — 行動裝置單欄堆疊，桌機三欄並排（TokNote · TokAgent · TokTodo）

Live Demo 立即體驗

https://tokai-pre-alpha-tokai.vercel.app/

TokAgent requires your own Anthropic API key. You can get one free at console.anthropic.com. Your key is stored locally in your browser and never saved on our servers.

TokAgent 需要您自備 Anthropic API 金鑰，可至 console.anthropic.com 免費獲取。金鑰僅儲存在您的瀏覽器本機，不會被我們的伺服器儲存。

Tech Stack

Layer Technology
Frontend React 18, TypeScript, Vite
Charts Recharts
Icons Lucide React
Fonts Share Tech Mono, Rajdhani, Inter (Google Fonts)
AI Anthropic Claude (claude-sonnet-4-5)
API Server Node.js, Express 5
Monorepo pnpm workspaces
Deployment Vercel (frontend + serverless API)

Architecture

Tokai/
├── artifacts/
│   ├── tokai/                  # React/Vite frontend
│   │   ├── src/
│   │   │   ├── pages/
│   │   │   │   └── dashboard.tsx   # Main dashboard
│   │   │   └── components/
│   │   │       └── agent-chat.tsx  # TokAgent chat UI
│   │   ├── public/
│   │   │   └── tokai_logo.png
│   │   └── vite.config.ts
│   └── api-server/             # Express.js API (serverless on Vercel)
│       └── api/
│           └── index.js        # /api/chat and /api/generate-description endpoints
├── pnpm-workspace.yaml
└── README.md

The frontend proxies /api requests to the API server in development. In production, VITE_API_BASE_URL points to the deployed Vercel serverless function.

Getting Started

Prerequisites

1. Clone the repository

git clone https://github.com/TokaiApp/Tokai.git
cd Tokai

2. Install dependencies

pnpm install

3. Configure environment variables

Frontend — create artifacts/tokai/.env.local:

# Leave empty for local dev (Vite proxy handles /api routing)
VITE_API_BASE_URL=

API server — create artifacts/api-server/.env:

# Optional fallback if users don't provide their own key
ANTHROPIC_API_KEY=sk-ant-...

4. Run the development servers

In two separate terminals:

# Terminal 1 — API server (port 3000)
PORT=3000 pnpm --filter @workspace/api-server dev

# Terminal 2 — Frontend (port 5173)
pnpm --filter @workspace/tokai dev

Open http://localhost:5173.

Deployment

The project deploys as two separate Vercel projects.

Frontend (artifacts/tokai/)

  1. Import artifacts/tokai/ as a Vercel project
  2. Framework preset: Vite
  3. Add environment variable: VITE_API_BASE_URL → your API server URL (e.g. https://your-api.vercel.app)

API Server (artifacts/api-server/)

  1. Import artifacts/api-server/ as a separate Vercel project
  2. No framework preset needed — the vercel.json configures it as a serverless function
  3. Optionally add ANTHROPIC_API_KEY as a fallback (users can also provide their own key in the UI)

Neural Metrics Explained

Metric Description
Focus Index Composite score (0–100) derived from EEG theta/beta wave patterns
Bio Energy Simulated biological energy level (%) — will reflect HRV and other biometric signals in future versions
Neural Noise Background EEG signal noise in μV² — lower is cleaner, higher values indicate distraction or arousal
T/B Ratio (Theta/Beta) Theta-to-beta wave ratio; elevated TBR (>3.0) is associated with ADHD inattention
Focus Window Predicted time remaining in your current focus state, based on recent trend

TokAgent uses all five values in its system prompt to tailor task planning recommendations:

  • High focus (>70): deep work, complex problem-solving, demanding cognitive tasks
  • Moderate focus (40–70): structured tasks, planning, communication, reviewing
  • Low focus (<40): easy wins, breaks, physical movement, administrative tasks

Data Privacy & Neuroprivacy

Brain data is among the most sensitive data a person can produce. Tokai is designed from the ground up to keep it under the user's control.

What data leaves your device — and where it goes

When you send a message to TokAgent, the following is transmitted to Anthropic's API:

  • Your current neural metrics (Focus Index, Bio Energy, Neural Noise, T/B Ratio)
  • Your TokTodo task list (titles, descriptions, demand tags, completion status)
  • Your TokNote journal entries (text, timestamps, mood tags, focus index at time of writing)
  • The conversation history for that session

Nothing else is transmitted. Your API key, your neural data, your tasks, and your journal entries are never sent to or stored on Tokai's own servers. Tokai's API server is a stateless relay — it receives the request, forwards it to Anthropic, and returns the response. It retains nothing.

Anthropic's no-training policy

Anthropic explicitly does not use API customer data to train its models. This is a documented commitment that distinguishes the API from consumer products. You can verify this in Anthropic's privacy policy.

Local-first storage

All user data — tasks, journal entries, API key, chat history — is stored exclusively in your browser's localStorage. It never leaves your device except via the TokAgent request described above. Clearing your browser data permanently deletes everything.

Open source auditability

The full source code, including the API server and the exact system prompt sent to Claude, is publicly available in this repository. There are no hidden data flows.

Alignment with the NeuroRights Foundation's Five Ethical Neurorights

Neuroright Tokai's implementation
Mental Privacy BYOK architecture; local-only storage; Anthropic's no-training API policy; open-source auditability
Personal Identity No writing to the brain; TokAgent is strictly advisory and cannot alter your state or identity
Free Will / Agency TokAgent cannot add, complete, or delete tasks — every action requires explicit user input
Fair Access Web-based, subscription-free, open source, BYOK (free Anthropic tier available), bilingual (EN / 繁中)
Algorithmic Bias Protection No demographic data collected; recommendations based solely on the user's own metrics; system prompt is publicly auditable

Known limitations

  • Shared devices: localStorage is device-local but not encrypted. If your device is shared or compromised, journal entries and tasks could be exposed. We recommend using a private browser profile.
  • Real EEG (future): When real EEG hardware integration ships, the privacy architecture will be re-evaluated. Raw EEG signals are significantly more sensitive than the simulated values used in this pre-alpha.
  • Anthropic dependency: TokAgent data handling is governed by Anthropic's privacy policy for the duration of each API call. Tokai does not control Anthropic's internal processing pipeline beyond what Anthropic publicly commits to.

Roadmap

  • Real EEG integration — Muse 2, OpenBCI, Neurosity
  • User accounts — persistent sessions, task history, focus analytics
  • ADHD-specific profiles — personalized thresholds and recommendations
  • Mobile app — native iOS/Android with EEG Bluetooth pairing
  • HRV and biometric integration — Apple Watch, Fitbit, Garmin
  • Focus session analytics — daily/weekly cognitive performance reports
  • Team/classroom mode — aggregate anonymous neural state for collaborative environments
  • LUNA integration — binary classification model for ADHD detection (research phase)

Research Context

Tokai originated as a master's degree thesis project exploring the intersection of real-time neurofeedback, agentic AI, and ADHD management. The core hypothesis: if an AI assistant has access to a user's live cognitive state, it can dramatically improve task planning outcomes for people with executive function challenges.

This repository represents the first public prototype (v0.1.0). We are actively seeking collaborators, researchers, and neurodivergent users willing to provide feedback.

Contributing

We welcome contributions — especially from people with ADHD or neurodiversity research backgrounds.

  1. Fork the repository
  2. Create a feature branch: git checkout -b feature/your-feature
  3. Commit your changes
  4. Open a pull request

For significant changes, please open an issue first to discuss the approach.

Feedback and bug reports: GitHub Issues

We especially want to hear from people who have tried the app. If you have ADHD, your experience matters most — this is built for you. 我們特別期待 ADHD 使用者的回饋，這款產品正是為你們而打造的。

About 關於

Tokai — Theory of Knowledge, Amplified Intelligence. Tokai — 知識理論，增強智能。

Learn more about the project and team at tokai.app.

License

Copyright © 2025 TokaiApp

Licensed under the Apache License, Version 2.0.

You may use, modify, and distribute this software freely under the terms of the Apache 2.0 License. See the LICENSE file for details.

Acknowledgments

  • Anthropic — Claude AI powering TokAgent
  • Recharts — charting library
  • Lucide — icons
  • The ADHD and neurodiversity community — for inspiring this work

About

tokai-pre-alpha-api-server.vercel.app

Resources

Readme

License

View license
Activity
Custom properties

Stars

1 star

Watchers

0 watching

Forks

0 forks
Report repository

Releases

1 tags

Packages

 
 
 

Contributors

Languages