365 Open Source Plan #006 · Local AI chat hub — one interface for all your providers
A local AI chat hub that doubles as an AI toolbox. One interface for all your AI providers, one click to start.
AI as a tool, not just a chatbot. Most AI interfaces are built for multi-turn conversations. But many tasks don't need context — you just want to throw text at an AI and get a result. Text polishing, translation, summarization, code review... you do these dozens of times a day, and previous conversations only get in the way.
ChatDeck's Stateless mode strips all history from every request. Each message is processed from scratch with only your system prompt — no context leaking in, no pollution from previous turns. Create a "Polish Text" profile, and it becomes a dedicated text-polishing tool: input in, result out, clean every time.
Combined with Profile presets (provider + system prompt, ready to go) and Compare mode (same input, multiple providers side-by-side), ChatDeck turns your local AI tools into a fast, organized workflow.
- Stateless mode - Zero context per message. Each request = system prompt + your input. Ideal for text editing, translation, and any high-frequency single-turn task
- Profile presets - Pre-configure provider + system prompt. Click to use, no setup each time
- Compare mode - Send the same prompt to multiple providers, pick the best result
- Free to use - Calls local CLI tools (
claude,qwen,ollama) directly, uses your existing subscriptions. API keys optional - 7 providers - Claude, OpenAI, Gemini, Qwen, DeepSeek, Groq, Ollama
Prerequisites: Node.js >= 18
npm install
npm run devOr use startup scripts: start.bat (Windows) / bash start.sh (macOS/Linux).
- Create a Profile - Pick a provider, set a system prompt, give it a name. Stateless mode is on by default
- Click and Chat - Select the profile from the sidebar, just type your message
- Come Back Anytime - All sessions are saved under each profile, easy to find and continue
- One click to chat - Pick a profile, start talking. No setup, no commands
- Stateless mode - Enabled per profile or toggled per session. Perfect for repetitive text tasks
- Compare mode - Side-by-side provider comparison
- 7 providers - Claude, OpenAI, Gemini, Qwen, DeepSeek, Groq, Ollama
- Free via CLI - Uses locally installed tools, no API cost
- Session fork - Edit any message and branch into a new conversation
- Export - Download conversations as Markdown
- Dark mode and Chinese/English interface
ChatDeck works in two ways per provider:
| Mode | How | Cost |
|---|---|---|
| CLI (recommended) | Uses locally installed tools like claude, ollama, qwen |
Free with your subscription |
| API | Direct API calls with your key | Pay per token |
CLI tools are auto-detected. To configure API keys, click the gear icon in the top-right corner.
| Provider | Where to get the key |
|---|---|
| Claude | console.anthropic.com (optional — works free via claude CLI) |
| Gemini | aistudio.google.com/apikey |
| Qwen | bailian.console.aliyun.com (DashScope API) |
| DeepSeek | platform.deepseek.com |
| Groq | console.groq.com |
| Ollama | No key needed (local) |
| Shortcut | Action |
|---|---|
Ctrl+N |
New session |
Ctrl+/ |
Toggle Compare mode |
Ctrl+E |
Export as Markdown |
Ctrl+1~4 |
Switch provider |
npm run build
npm startOr via CLI:
npx chat-deck # default port 3456
npx chat-deck -p 8080 # custom portpackages/
client/ React + Vite frontend
server/ Express + WebSocket backend
shared/ Shared TypeScript types
This is project #006 of the 365 Open Source Plan.
One person + AI, 300+ open source projects in a year. Submit your idea →
MIT

