Lite SuiteLite Suite

Lite Suite

The AI operating environment — infinite canvas workspace with 29 panel types, voice pipeline, agent orchestration, and local-first AI

Overview

Lite Suite is a desktop AI operating environment built as a single Electron application. Code editors, terminals, AI agents, browsers, benchmarks, voice control, memory systems, and design tools all live together as composable panels on an infinite canvas. Everything runs locally.

The early product line split these workflows into focused tools. The post-merge architecture brought the core workflows back into LiteSuite as built-in panels and services, while LiteImage and LiteModeler remain standalone GPU studios and LiteDock remains the free Windows utility.

Architecture

LiteSuite (Electron AI operating environment, Bun monorepo)
├── apps/desktop/        # Electron main — windows, IPC, services
├── apps/server/         # Bun/Effect backend — WebSocket, SQLite
├── apps/web/            # React 19 renderer — panels, canvas, stores
├── packages/contracts/  # TypeScript schemas, WS protocol
├── packages/shared/     # Net service, logging, utilities
├── resources/litesuite/ # 79 agents, 29 skills, commands, plugins
└── resources/mcp-server/# MCP tools (21 tools)

Workspace Modes

Canvas Mode

An infinite 2D surface. Pan, zoom, and arrange panels anywhere. Multiple panels visible simultaneously — code on the left, terminal in the center, AI chat on the right. Spatial arrangement reflects how you think about your project.

Zen Mode

Single focused panel, no distractions. Switch between panels via the sidebar or keyboard shortcuts.

Panel System

29 panel types across four categories:

| Category | Panels | |----------|--------| | Workspace | Terminal, Unified Editor, Browser, Files, Git | | AI | Frontier Chat, Claude, Codex, LiteAgent Chat, Claude Teams | | Tools | Benchmark, Arena, YouTube, Voice, Memory Graph, Model Hub | | System | Dashboard, Agent, Settings, Design System, Screens |

Add panels via the + button in the sidebar or Ctrl+Shift+N.

Voice Pipeline

Built-in voice system with:

  • STT — Local Whisper transcription
  • TTS — Qwen3-TTS with emotion classification and 11 visualizer modes
  • Wake detection — hands-free activation
  • 48 voice commands — navigate panels, control the workspace, trigger actions
  • Companion mode — full voice conversation with an AI that remembers context

See Voice Pipeline for details.

Agent Orchestration

Lite Suite ships with the LiteHarness — a 5-tier agent orchestration system:

| Tier | Role | |------|------| | Sentinel | Watches all agents, intervenes when needed | | Orchestrator | Plans and coordinates multi-agent workflows | | Leader | Manages a team of workers | | Worker | Executes individual tasks | | Thinker | Read-only analysis and review |

79 canonical agents (40 polymathic thinking agents, 14 thinkers, 9 PRD generators, 6 pen testers, 5 specialists, 5 meta-agents) ship with the app.

MCP Tools

21 MCP tools built in, covering browser automation (9 tools), filesystem, process management, agent bridge, and terminal multiplexing. See MCP Servers for the full catalog.

Theming

9 preset themes with Abyss (ultra-dark) as the default. 17 color tokens, EmberSparks particle system, glass blur effects, and accent color picker. Appearance is configured in Settings.

Licensing

Machine-tied HMAC-SHA256 license keys activated via litesuite:// deep links. 7-day trial with full functionality, then Pro tier. License validation against the litesuite.dev API.

Tech Stack

| Component | Technology | |-----------|-----------| | Runtime | Electron 40 | | Package Manager | Bun | | Frontend | React 19, Zustand, Tailwind CSS 4 | | Backend | Bun, Effect-TS | | Persistence | SQLite | | Protocol | WebSocket (port 3773) | | Icons | Lucide |

Port Map

| Port | Service | |------|---------| | 3773 | Backend server (HTTP + WebSocket) | | 7423 | Agent Bridge (REST API) | | 7438 | Voice API server | | 5123 | TTS Server (Qwen3, on-demand) | | 8080 | Whisper STT (on-demand) |

Requirements

  • OS: Windows 10/11 (64-bit)
  • RAM: 16 GB minimum, 32+ GB recommended
  • GPU: NVIDIA GPU recommended for local AI models
  • Disk: ~2 GB base, more for local models