GitPedia

Devonz

Full-stack vibe coding platform with model freedom. Any LLM, local or cloud, zero lock-in.

From zebbern·Updated June 24, 2026·View on GitHub·

**AI-powered vibe coding platform — describe what you want, watch it build.** The project is written primarily in TypeScript, first published in 2026. Key topics include: ai-generator, ai-powered, ai-website-builder, ai-website-generator, backend-development.

Latest release: v1.1.0— v1.1.0 - Stable Release
February 19, 2026View Changelog →
<div align="center"> <img width="236" height="79" alt="devonz" src="https://github.com/user-attachments/assets/30c464d9-39a9-4c0d-85f8-64473cfa774c" />

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

</div>

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

RequirementVersion
Node.js≥ 18.18.0
pnpm9.x (see below)

Enable pnpm via Corepack:

bash
corepack enable
corepack prepare pnpm@9.14.4 --activate

Setup

bash
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:

bash
cp .env.example .env.local

Or run the interactive setup wizard:

bash
pnpm run setup

At minimum, add one AI provider key:

bash
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

ProviderEnv VariableNotes
OpenAIOPENAI_API_KEYGPT-4o, o1 series
AnthropicANTHROPIC_API_KEYClaude 3.5 / 4
Google GeminiGOOGLE_GENERATIVE_AI_API_KEYGemini models
xAIXAI_API_KEYGrok models
GroqGROQ_API_KEYFast inference
MistralMISTRAL_API_KEYMistral Large / Small
DeepSeekDEEPSEEK_API_KEYDeepSeek Chat, Coder
CohereCOHERE_API_KEYCommand R+
PerplexityPERPLEXITY_API_KEYSearch-augmented
TogetherTOGETHER_API_KEYOpen-source models
FireworksFIREWORKS_API_KEYFast inference
CerebrasCEREBRAS_API_KEYFast inference
HuggingFaceHuggingFace_API_KEYInference API
OpenRouterOPEN_ROUTER_API_KEYMulti-provider routing
GitHub ModelsGITHUB_API_KEYGitHub-hosted models
MoonshotMOONSHOT_API_KEYKimi models
HyperbolicHYPERBOLIC_API_KEYHyperbolic inference
Z.aiZAI_API_KEYGLM coding models
Amazon BedrockAWS_BEDROCK_CONFIGAWS-hosted models (JSON)
OllamaOLLAMA_API_BASE_URLLocal models, no key needed
LM StudioLMSTUDIO_API_BASE_URLLocal models, no key needed
OpenAI-LikeOPENAI_LIKE_API_BASE_URLAny 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:

bash
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.

bash
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

bash
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

ProfileCommandDescription
(default)docker compose up -dProduction — pulls from GHCR
devdocker compose --profile dev upDevelopment — bind-mount with hot reload
auto-updatedocker compose --profile auto-update up -dAdds Watchtower for automatic image updates

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

Tech Stack

LayerTechnology
FrameworkReact Router v7 (Remix)
UIReact 19
BuildVite 5
LanguageTypeScript 5 (strict)
StylingUnoCSS + SCSS + Radix UI
DatabaseDrizzle ORM
StateNanostores
AIVercel AI SDK (22 providers)
RuntimeWebContainer API
EditorCodeMirror 6
Terminalxterm.js
TestingVitest + Testing Library + Playwright
Package Managerpnpm 9.14.4

Development

Scripts

bash
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

bash
# 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

<details> <summary>Expand file tree</summary> 
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
</details>

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

<div align="center"> <strong>Build anything with AI. Just describe what you want.</strong> <br><br> <a href="https://github.com/zebbern/Devonz">GitHub</a> · <a href="https://github.com/zebbern/Devonz/issues">Issues</a> · <a href="docs/">Documentation</a> </div>

Contributors

Showing top 12 contributors by commit count.

zebbern

zebbern

586 commits

thecodacus

thecodacus

320 commits

dustinwloring1988

dustinwloring1988

205 commits

Stijnus

Stijnus

163 commits

xKevIsDev

xKevIsDev

163 commits

github-actions[bot]

github-actions[bot]

129 commits

wonderwhy-er

wonderwhy-er

95 commits

coleam00

coleam00

85 commits

d3lm

d3lm

42 commits

leex279

leex279

26 commits

kc0tlh

kc0tlh

25 commits

dependabot[bot]

dependabot[bot]

20 commits

View all contributors on GitHub →

This article is auto-generated from zebbern/Devonz via the GitHub API.Last fetched: 6/25/2026