@uniqueli/openwork

A desktop interface for deepagentsjs — an opinionated harness for building deep agents with filesystem capabilities, planning, and subagent delegation.

✨ Enhanced with Multiple Custom API Support - Add unlimited OpenAI-compatible API providers with a single click!

[!CAUTION] openwork gives AI agents direct access to your filesystem and the ability to execute shell commands. Always review tool calls before approving them, and only run in workspaces you trust.

Get Started

# Run directly with npx
npx @uniqueli/openwork

# Or install globally
npm install -g @uniqueli/openwork
openwork

Requires Node.js 18+.

From Source

Supported Models

| Provider | Models | | --------- | ----------------------------------------------------------------- | | Anthropic | Claude Opus 4.5, Claude Sonnet 4.5, Claude Haiku 4.5, Claude Opus 4.1, Claude Sonnet 4 | | OpenAI | GPT-5.2, GPT-5.1, o3, o3 Mini, o4 Mini, o1, GPT-4.1, GPT-4o | | Google | Gemini 3 Pro Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash Lite | | Custom | Add unlimited custom providers! |

✨ Multiple Custom API Providers

New in v0.2.0: Add multiple custom OpenAI-compatible API providers directly from the UI!

How to Add Custom Providers

1. Click the model selector in the chat interface
2. Click the "+ 添加Provider" button at the bottom of the provider list
3. Fill in the form:
   • ID: Unique identifier (e.g., moonshot, zhipu, deepseek)
   • Display Name: Name shown in UI (e.g., Moonshot AI, Zhipu AI)
   • Base URL: API endpoint (e.g., https://api.moonshot.cn/v1)
   • API Key: Your API key
   • Model Name: Model identifier (e.g., kimi-k2-turbo-preview)
4. Click Save - your new provider appears immediately!

Supported Custom APIs

Works with any OpenAI-compatible API:

• Chinese AI Providers: Moonshot AI (Kimi), Zhipu AI (GLM), DeepSeek, Baichuan, etc.
• Self-hosted models: vLLM, Text Generation WebUI, LocalAI, Ollama (with OpenAI compatibility)
• Cloud services: Azure OpenAI, AWS Bedrock (with proxy), Cloudflare AI
• Other providers: Together AI, Anyscale, Fireworks AI, etc.

Example Configurations

Moonshot AI (Kimi)

ID: moonshot
Display Name: Moonshot AI
Base URL: https://api.moonshot.cn/v1
Model Name: kimi-k2-turbo-preview

Zhipu AI (GLM)

ID: zhipu
Display Name: Zhipu AI
Base URL: https://open.bigmodel.cn/api/paas/v4
Model Name: glm-4-plus

DeepSeek

ID: deepseek
Display Name: DeepSeek
Base URL: https://api.deepseek.com/v1
Model Name: deepseek-chat

Configure via Settings UI or by setting environment variables:

CUSTOM_BASE_URL=https://api.example.com/v1
CUSTOM_API_KEY=your-api-key
CUSTOM_MODEL=your-model-name  # optional

See CUSTOM_API.md for detailed instructions.

Changelog

v0.2.1 (2026-01-19)

• 🐛 Critical Fix: Fixed "Missing credentials" error for users without OpenAI API key
• 🔧 Custom API now works correctly even when OPENAI_API_KEY is not set in environment
• 📝 Improved logging for debugging custom API configurations

v0.2.0 (2026-01-18)

• ✨ Multiple Custom API Providers: Add unlimited custom providers via UI
• 🎨 Improved UX: One-click provider addition with "+ 添加Provider" button
• 🔧 Better Configuration: Each provider has its own name, base URL, API key, and model
• 🌐 Chinese AI Support: Perfect for Moonshot AI, Zhipu AI, DeepSeek, and other providers
• 📝 Simplified Settings: Cleaner settings dialog focused on standard providers

v0.1.0 (2026-01-15)

• 🎉 Initial release with basic custom API support
• 🔑 Single custom API configuration via Settings

Contributing

We welcome contributions! See CONTRIBUTING.md for guidelines.

Report bugs via GitHub Issues.

Credits

This project is a fork of openwork by LangChain with enhanced custom API support.

License

MIT — see LICENSE for details.