The desktop client that makes AI agents accessible to everyone.
Powered by Hermes Agent from Nous Research.
Running AI agents today means wrestling with Python environments, terminal commands, and config files. OpenOtter removes all of that. Download the app, enter your API key, and start chatting with a fully autonomous agent — complete with tool use, memory, and multi-platform deployment.
- Zero-config setup — One-click install on macOS and Windows. No Python, no pip, no terminal.
- Visual agent management — Create, configure, and monitor multiple agents from a clean GUI.
- Multi-provider support — Seamlessly switch between 10+ LLM providers, including Chinese providers.
- Deploy anywhere — Connect your agents to Feishu, DingTalk, WeCom, WeChat, QQ Bot, and more.
- Built for power users too — Credential pooling, cron scheduling, MCP servers, and session search.
- Setup Wizard — Guided first-run experience with automatic Hermes Agent installation
- Agent Creation — Template-based wizard with persona (SOUL.md), model selection, and platform binding
- Dashboard — Real-time overview of running agents, session stats, and system insights
- Chat Interface — Markdown rendering, code highlighting, and streaming responses
- IM Channels — Deploy agents to Feishu (飞书), DingTalk (钉钉), WeCom (企业微信), WeChat (微信), QQ Bot, Popo, NetEase Bee, NIM
- Model Providers — OpenRouter, OpenAI, Anthropic, DeepSeek, Zhipu (GLM), DashScope (Qwen), Kimi, MiniMax, Groq, and custom endpoints
- Credential Pool — Manage multiple API keys per provider with rotation support
- MCP Servers — Discover and manage Model Context Protocol tool servers
- Cron Jobs — Schedule recurring agent tasks
- Skills Management — Browse and manage agent skill files
- Memory & Sessions — Search through agent conversation history and memory
- Dark / Light Theme — Polished UI with system-level theme toggling
- Auto-update — Built-in update mechanism for seamless upgrades
- China Network Detection — Automatic mirror selection for users behind the GFW
| Provider | Models |
|---|---|
| OpenRouter | 200+ models (GPT, Claude, Llama, Mistral, etc.) |
| OpenAI | GPT-4.1, o3, o4-mini |
| Anthropic | Claude Sonnet 4, Claude Opus 4 |
| DeepSeek | DeepSeek-V3, DeepSeek-R1 |
| Zhipu AI | GLM-4, GLM-4-Plus |
| DashScope | Qwen-Max, Qwen-Plus |
| Kimi | Moonshot-v1 |
| MiniMax | abab6.5s |
| Groq | Llama 3, Mixtral |
| Custom | Any OpenAI-compatible endpoint |
Coming soon — pre-built binaries for macOS (.dmg) and Windows (.exe).
git clone https://github.com/your-org/openotter.git
cd openotter
npm install
npm run sidecar:dev # Create dev sidecar wrapper
npm run tauri dev # Start developmentnpm run sidecar:build # Build sidecar binary (requires hermes-agent)
npm run tauri build # Build production appopenotter/
├── src/ # React frontend (TypeScript)
│ ├── components/ # UI: Dashboard, Chat, Agents, Channels, Settings, ...
│ ├── stores/ # Zustand state management
│ └── lib/ # Hermes bridge (Tauri IPC bindings)
├── src-tauri/ # Rust backend
│ ├── src/ # Tauri commands, sidecar, gateway, installer
│ ├── binaries/ # Sidecar binaries (gitignored)
│ └── capabilities/ # Tauri v2 permissions
├── public/ # Static assets (logo, platform icons)
├── scripts/ # Build scripts (PyInstaller sidecar)
└── .github/workflows/ # CI/CD (GitHub Actions)
| Layer | Technology |
|---|---|
| Desktop Shell | Tauri v2 (Rust) |
| Frontend | React 19 + TypeScript + Tailwind CSS v4 |
| State | Zustand |
| Bundler | Vite 7 |
| Agent Engine | Hermes Agent (Python, managed as sidecar) |
Contributions are welcome! Please open an issue to discuss your idea before submitting a PR.