Skip to content

zebbern/Devonz

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2,230 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

devonz

AI-powered vibe coding platform — describe what you want, watch it build.

Node TypeScript License: MIT

Features · Quick Start · Environment Variables · Docker · Tech Stack · Contributing

Devonz Screenshot


What is Devonz?

Devonz is a local AI-powered vibe coding platform forked from bolt.diy. Describe what you want to build in natural language, and an AI agent generates full-stack applications in a browser-based IDE with an integrated editor, terminal, and live preview.

This is a local development tool, not a hosted web service. You run it on your own machine, bring your own API keys, and keep full control of your code.


Features

  • Real-time code generation — AI streaming produces code as you watch, with live diffs applied to the editor.
  • In-browser code execution — run and preview generated applications directly in the browser via WebContainer.
  • 22+ LLM providers — OpenAI, Anthropic, Google, xAI, Groq, Mistral, DeepSeek, OpenRouter, Ollama, LM Studio, and more. Swap models mid-conversation.
  • Agent mode — handles complex multi-step tasks with autonomous planning and execution.
  • Template system — start from React, Astro, Vite + Shadcn, and other popular boilerplates, then customise with AI.
  • Git integration — built-in version control with commit, branch, and push support.
  • Smart terminal — integrated terminal with error detection and intelligent alerts that help the agent auto-fix failures.
  • Speech-to-text — dictate prompts instead of typing.
  • File upload — attach files and images to prompts for additional context.
  • Dark / light themes — system-aware theme switching.
  • Framer Motion animations — smooth chat animations and transitions.
  • Error classification — errors are categorised and routed to toast notifications or dialog windows based on severity.
  • Deploy from the UI — push to GitHub, GitLab, Netlify, or Vercel directly.

Quick Start

Prerequisites

Requirement Version
Node.js ≥ 18.18.0
pnpm 9.x (see below)

Enable pnpm via Corepack:

corepack enable
corepack prepare pnpm@9.14.4 --activate

Setup

git clone https://github.com/zebbern/Devonz.git
cd Devonz
cp .env.example .env.local      # add at least one API key
pnpm install
pnpm run dev

Open http://localhost:5173.

The first load can take up to two minutes while dependencies compile.


Environment Variables

Copy the example file and fill in your API keys:

cp .env.example .env.local

Or run the interactive setup wizard:

pnpm run setup

At minimum, add one AI provider key:

OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=sk-ant-...
GOOGLE_GENERATIVE_AI_API_KEY=AI...

You can also configure providers at runtime via Settings → Providers in the UI.

See .env.example for the full list of supported variables including deployment tokens, local provider URLs, and encryption settings.

Supported LLM Providers

Provider Env Variable Notes
OpenAI OPENAI_API_KEY GPT-4o, o1 series
Anthropic ANTHROPIC_API_KEY Claude 3.5 / 4
Google Gemini GOOGLE_GENERATIVE_AI_API_KEY Gemini models
xAI XAI_API_KEY Grok models
Groq GROQ_API_KEY Fast inference
Mistral MISTRAL_API_KEY Mistral Large / Small
DeepSeek DEEPSEEK_API_KEY DeepSeek Chat, Coder
Cohere COHERE_API_KEY Command R+
Perplexity PERPLEXITY_API_KEY Search-augmented
Together TOGETHER_API_KEY Open-source models
Fireworks FIREWORKS_API_KEY Fast inference
Cerebras CEREBRAS_API_KEY Fast inference
HuggingFace HuggingFace_API_KEY Inference API
OpenRouter OPEN_ROUTER_API_KEY Multi-provider routing
GitHub Models GITHUB_API_KEY GitHub-hosted models
Moonshot MOONSHOT_API_KEY Kimi models
Hyperbolic HYPERBOLIC_API_KEY Hyperbolic inference
Z.ai ZAI_API_KEY GLM coding models
Amazon Bedrock AWS_BEDROCK_CONFIG AWS-hosted models (JSON)
Ollama OLLAMA_API_BASE_URL Local models, no key needed
LM Studio LMSTUDIO_API_BASE_URL Local models, no key needed
OpenAI-Like OPENAI_LIKE_API_BASE_URL Any compatible API

See docs/LLM-PROVIDERS.md for details on configuring each provider.

Encryption

Set DEVONZ_ENCRYPTION_KEY to persist encrypted API-key cookies across server restarts. Generate a key with:

node -e "console.log(require('crypto').randomBytes(32).toString('hex'))"

If omitted, a random key is generated on each startup and cookies will not survive restarts.


Docker

No local Node.js required — just Docker.

git clone https://github.com/zebbern/Devonz.git
cd Devonz
cp .env.example .env.local      # add at least one API key
docker compose up -d

Open http://localhost:5173.

Docker Commands

pnpm docker:build          # Build image locally
pnpm docker:run            # Run standalone container
pnpm docker:up             # Start via Docker Compose
pnpm docker:down           # Stop services
pnpm docker:dev            # Dev mode with hot reload
pnpm docker:update         # Pull latest image + restart

Compose Profiles

Profile Command Description
(default) docker compose up -d Production — pulls from GHCR
dev docker compose --profile dev up Development — bind-mount with hot reload
auto-update docker compose --profile auto-update up -d Adds Watchtower for automatic image updates

Set DEVONZ_ENCRYPTION_KEY in your environment or .env.local to persist encrypted cookies across container restarts.


Tech Stack

Layer Technology
Framework React Router v7 (Remix)
UI React 19
Build Vite 5
Language TypeScript 5 (strict)
Styling UnoCSS + SCSS + Radix UI
Database Drizzle ORM
State Nanostores
AI Vercel AI SDK (22 providers)
Runtime WebContainer API
Editor CodeMirror 6
Terminal xterm.js
Testing Vitest + Testing Library + Playwright
Package Manager pnpm 9.14.4

Development

Scripts

pnpm run dev               # Start dev server (http://localhost:5173)
pnpm run build             # Production build
pnpm run start             # Run production build
pnpm run preview           # Build + preview locally

pnpm test                  # Run test suite
pnpm test:watch            # Tests in watch mode
pnpm run typecheck         # TypeScript type check
pnpm run lint              # ESLint check
pnpm run lint:fix          # Auto-fix lint issues

pnpm run clean             # Clean build artifacts
pnpm run update            # Pull latest + install + rebuild

Updating

# From source
pnpm run update                    # pull, install, rebuild
pnpm run update -- --skip-build    # pull + install only

# Docker
pnpm docker:update                 # pull latest image, restart

Project Structure

Expand file tree
devonz/
├── app/
│   ├── components/         # React components
│   │   ├── @settings/      # Settings panel
│   │   ├── chat/           # Chat interface
│   │   ├── deploy/         # Deployment integrations
│   │   ├── editor/         # Code editor (CodeMirror)
│   │   ├── git/            # Git integration
│   │   ├── header/         # App header
│   │   ├── sidebar/        # Sidebar navigation
│   │   ├── templates/      # Template gallery
│   │   ├── ui/             # Shared UI components
│   │   └── workbench/      # Development workbench
│   ├── lib/
│   │   ├── .server/        # Server-only code (LLM streaming)
│   │   ├── agent/          # Agent prompts + orchestration
│   │   ├── common/         # Prompt library + shared prompts
│   │   ├── hooks/          # Custom React hooks
│   │   ├── modules/        # Feature modules (22 LLM providers)
│   │   ├── persistence/    # IndexedDB, localStorage, autoBackup
│   │   ├── runtime/        # Action runner, message parser
│   │   ├── services/       # API services
│   │   ├── stores/         # Nanostores
│   │   └── utils/          # Utilities
│   ├── routes/             # Routes (pages + API endpoints)
│   ├── styles/             # Global styles (SCSS + CSS)
│   └── types/              # TypeScript type definitions
├── docs/                   # Extended documentation
├── drizzle/                # Database migrations
└── scripts/                # Build & update scripts

Contributing

  1. Fork the repository
  2. Create a feature branch: git checkout -b feature/my-feature
  3. Run checks: pnpm lint:fix && pnpm test
  4. Commit with a conventional message: git commit -m 'feat: add my feature'
  5. Push and open a Pull Request

See docs/CONTRIBUTING.md for detailed guidelines including code style, testing patterns, and API route conventions.


Acknowledgements

Devonz is forked from bolt.diy (originally StackBlitz bolt.new). Built with the Vercel AI SDK.


License

MIT


Build anything with AI. Just describe what you want.

GitHub · Issues · Documentation