Cette page est affichée en anglais. Une traduction française est en cours.
ScriptsJul 29, 2026·3 min de lecture

Open Codesign — Open-Source Multi-Model AI Design Tool with BYOK

A local-first, multi-model design application that generates prototypes, slides, and PDFs from prompts, supporting Claude, GPT, Gemini, and local models via Ollama with MIT-licensed source code.

Prêt pour agents

Installation avec revue préalable

Cet actif nécessite une revue. Le prompt copié demande un dry-run, affiche les écritures, puis continue seulement après confirmation.

Needs Confirmation · 66/100Policy : confirmer
Surface agent
Tout agent MCP/CLI
Type
Skill
Installation
Single
Confiance
Confiance : Established
Point d'entrée
Open Codesign Guide
Commande avec revue préalable
npx -y tokrepo@latest install 6b3a8e27-8ae4-11f1-9bc6-00163e2b0d79 --target codex

Dry-run d'abord, confirmez les écritures, puis lancez cette commande.

Introduction

Open Codesign is an open-source desktop application for generating visual designs from text prompts using any LLM provider. It supports Claude, GPT, Gemini, Kimi, GLM, and local models through Ollama, all configured via Bring Your Own Key (BYOK). The app generates prototypes, slides, and PDF documents, running entirely on your local machine under the MIT license.

What Open Codesign Does

  • Generates UI prototypes, slide decks, and PDF documents from natural language descriptions
  • Supports multiple LLM providers including Claude, GPT, Gemini, Kimi, GLM, and Ollama
  • Uses a BYOK model so you connect your own API keys without any platform subscription
  • Renders designs locally with live preview and iterative refinement through follow-up prompts
  • Exports to production-ready formats including HTML, PDF, and image files

Architecture Overview

Open Codesign is an Electron desktop app with a TypeScript frontend. It uses a provider abstraction layer to communicate with different LLM APIs, sending design prompts and receiving structured UI descriptions. The rendering engine converts these descriptions into visual components using a templating system. All rendering happens locally in the Electron webview, and exported files are written directly to disk. The architecture keeps provider logic separate from the rendering pipeline, making it straightforward to add new LLM backends.

Self-Hosting & Configuration

  • Clone the repository and install dependencies with npm install
  • Start development mode with npm run dev or build production binaries with npm run build
  • Add your API keys for any supported provider in the settings panel
  • For local model support, install Ollama separately and configure the endpoint in settings
  • The app stores all data locally with no cloud dependency beyond LLM API calls

Key Features

  • Multi-model support with a unified interface across Claude, GPT, Gemini, and local models
  • BYOK approach means no platform fees beyond your own API costs
  • Local-first architecture ensures design data stays on your machine
  • MIT-licensed source code with full transparency and modification rights
  • Iterative design workflow with live preview and prompt refinement

Comparison with Similar Tools

  • Open Design — similar concept but larger feature set; Open Codesign focuses on simplicity and multi-model support
  • Claude Design — proprietary Anthropic product; Open Codesign is open-source and model-agnostic
  • v0 by Vercel — cloud-based UI generation; Open Codesign runs locally with any model
  • Figma AI — cloud design tool with AI features; Open Codesign is fully local and open-source
  • Bolt.new — AI app generator; Open Codesign focuses on design outputs like slides and PDFs rather than full apps

FAQ

Q: Can I use Open Codesign completely offline? A: Yes, if you use a local LLM via Ollama. Cloud providers require an internet connection for API calls.

Q: Which export formats are supported? A: HTML, PDF, and image formats. The export pipeline generates clean, production-ready files.

Q: Is there a cost to use Open Codesign? A: The application itself is free and MIT-licensed. You pay only for your own LLM API usage.

Q: How do I add support for a new LLM provider? A: The provider abstraction layer makes it straightforward to add new backends by implementing the provider interface in TypeScript.

Sources

Fil de discussion

Connectez-vous pour rejoindre la discussion.
Aucun commentaire pour l'instant. Soyez le premier à partager votre avis.

Actifs similaires