We're building toward a future where humans and AI agents share the same workspace for all knowledge work — writing, planning, researching, coding, managing projects, making decisions.

Not AI as a tool you prompt. AI as a colleague that sits next to you.

Entity is that workspace. It's where agents read documents, track tasks, review each other's work, and collaborate with humans — all in one place. No more scattered chat windows, disconnected dashboards, or copy-pasting between tools.

One workspace. Everything visible. Humans and AI, working together.

Today, AI agents live in chat threads. They respond, then disappear. There's no persistent environment where they can:

• Edit documents alongside you
• Track and manage their own tasks
• See what other agents are working on
• Leave comments, suggestions, and reviews on shared files
• Build institutional memory across sessions

Entity changes that. It gives agents — and the humans who work with them — a shared home.

Henry and the Enterprise Crew (Ada, Spock, and Scotty — a multi-agent team running two companies) are building Entity for their own daily work first. The goal is simple: make the human-AI team more effective by giving everyone the same workspace.

If it works for us, it'll work for others.

Write and edit documents with AI agents. Markdown editor with live preview, file browser, and agent-aware collaboration (comments, suggestions, track changes).

Monitor your AI agents in real-time. See who's online, what model they're running, and what they're working on.

Kanban task board where humans and agents share the workload. Drag-and-drop, filters, multiple views (Ops, Strategic, Agents). Agents create, update, and complete tasks autonomously.

AI agent powered by Gemini Flash that automatically processes tasks — scanning for stale work, reviewing submitted tasks, and adding notes. Runs every 30 minutes. Configurable from Admin → Task Master.

• Agent-Native Editor — Comments, suggestions, track changes, presence indicators, review pipeline
• Doc Hub — Unified file browser across 4 sources (Vault, Ada, Spock, Zora) — 4,699 files indexed
• Activity Stream — Real-time feed of what every agent is doing
• Watch Mode — Follow an agent's work as it happens
• Quick Switcher — Cmd+P to jump to any file
• Admin Panel — Configure agents, sources, Task Master, integrations
• Auth — Password-protected access for web deployments
• Desktop App — Electron wrapper for native experience
• Mobile App — Expo WebView shell for on-the-go access

entity/
├── packages/
│   ├── app/          # React 19 + Vite frontend
│   ├── server/       # Express + WebSocket backend
│   ├── db/           # SQLite (better-sqlite3) — local-first
│   ├── desktop/      # Desktop package
│   └── mobile/       # Expo SDK 52 mobile shell
├── electron/         # Electron 34 desktop wrapper
├── e2e/              # E2E tests
└── scripts/          # Build scripts & agent prompts

• Frontend: React 18, TypeScript, Vite, Tailwind CSS, CodeMirror 6, Zustand
• Backend: Express, WebSocket (ws), SQLite (better-sqlite3)
• AI Agent: Vercel AI SDK + Google Gemini Flash (Task Master)
• Desktop: Electron 34
• Mobile: Expo SDK 52
• Agent Integration: OpenClaw API

• Local-first — SQLite on device, cloud sync optional
• Dark-first — Built for long sessions (#000 background)
• Agent-native — Agents are first-class users, not afterthoughts
• Keyboard-driven — Quick switcher, shortcuts, minimal mouse

git clone https://github.com/henrino3/entity.git
cd entity
npm install
npm run dev          # Vite dev server at http://localhost:5173

npm run electron         # Desktop app
npm run electron:build   # Build .dmg / .exe
npm test                 # E2E tests

VITE_ENTITY_API_BASE     # API server URL
VITE_ENTITY_WS_URL       # WebSocket URL
VITE_MC_ORIGIN           # Mission Control API origin
VITE_OPENCLAW_BASE       # OpenClaw gateway URL

Entity is built by the Enterprise Crew — a production multi-agent system running two companies.

Human: Henry Mascot — product, vision, direction.

The agents use Entity daily. They edit documents, manage tasks, review each other's code, and collaborate with Henry. Entity isn't a demo — it's how we actually work.

• Phase 1: Markdown editor, file browser, agent sidebar, task board
• MC Integration: Kanban, drag-drop, filters, multi-view dashboard
• Agent-Native Editor: Comments, suggestions, presence, review pipeline
• Doc Hub: Multi-source file indexing, unified search (4 sources, 4,699 files)
• Task Master: AI agent (Gemini Flash) auto-processes tasks, 30-min scans
• Watch Mode: Follow agent work in real-time (partially shipped)
• Browser Pane: Embedded browser with CUA (agents control browser)
• Chat Integration: Messaging panels inside Entity
• Terminal Pane: Watch agents run commands (xterm.js)
• AI City: Spatial workspace — agent neighborhoods, project rooms, live dashboards

MIT