Show HN:Sim Studio——开源代理工作流GUI
Show HN: Sim Studio – Open-Source Agent Workflow GUI

原始链接: https://github.com/simstudioai/sim

Sim Studio 允许你构建、测试和优化自主代理工作流程。你可以使用云端托管版本,也可以使用 Docker 自行托管。推荐的自托管方法包括克隆仓库,使用必要的变量(例如 `BETTER_AUTH_SECRET`)配置 `.env` 文件,然后运行 `docker compose up -d --build`。 对于本地模型的使用,请使用 `ollama_docker.sh` 拉取模型,然后使用相应的 `docker compose` 命令(根据你的硬件使用 `--profile local-gpu` 或 `--profile local-cpu`)启动支持本地模型的 Sim Studio。 或者,在克隆仓库后,使用 VS Code 和 Remote - Containers 扩展直接进行开发。使用 `npm install` 安装依赖项,配置 `.env` 文件(设置数据库和身份验证),使用 `npx drizzle-kit push` 推送数据库模式,并使用 `npm run dev` 启动开发服务器。记住在生产环境中正确配置 `RESEND_API_KEY`。访问应用程序的地址为 http://localhost:3000。

Sim Studio是由Emir和Waleed构建的开源平台,它提供了一个可视化的、拖放式的界面,用于创建和管理多智能体工作流程。Sim Studio解决了构建可靠的智能体系统所面临的复杂性问题,允许用户在一个有向图中定义智能体的交互、工具的使用和逻辑流程。其主要功能包括预构建的集成、细粒度的工具使用控制、条件块和基于LLM的路由。 与现有的框架不同,Sim Studio提供了对智能体行为的显式控制并减少了抽象,从而提高了透明度并加快了迭代速度。该平台还支持工作流程模拟、API部署、Webhook和内置的可观察性功能,用于性能监控。用户可以使用它构建各种应用程序,例如欺诈检测代理和人力资源聊天机器人。Sim Studio采用Apache 2.0许可证,并欢迎社区贡献。团队相信其以工作流程为中心的方法使智能体开发更加易于访问和可靠。

原文

Sim Studio Logo

License: Apache-2.0 Discord Twitter PRs welcome support

Sim Studio is a powerful, user-friendly platform for building, testing, and optimizing agentic workflows.

  1. Run on our cloud-hosted version
  2. Self-host

There are several ways to self-host Sim Studio:

Option 1: Docker Environment (Recommended)

# Clone your forked repository
git clone https://github.com/YOUR_USERNAME/sim.git
cd sim

# Create environment file and update with required environment variables (BETTER_AUTH_SECRET)
cp sim/.env.example sim/.env

# Start Sim Studio using the provided script
docker compose up -d --build

or

./start_simstudio_docker.sh

After running these commands:

  1. Access the Application:

  2. Useful Docker Commands:

    # View application logs
    docker compose logs -f simstudio
    
    # Access PostgreSQL database
    docker compose exec db psql -U postgres -d simstudio
    
    # Stop the environment
    docker compose down
    
    # Rebuild and restart (after code changes)
    docker compose up -d --build

Working with Local Models

To use local models with Sim Studio, follow these steps:

  1. Pull Local Models

    # Run the ollama_docker.sh script to pull the required models
    ./sim/scripts/ollama_docker.sh pull <model_name>
  2. Start Sim Studio with Local Models

    #Start Sim Studio with local model support
    ./start_simstudio_docker.sh --local
    
    # or
    
    # Start Sim Studio with local model support if you have nvidia GPU
    docker compose up --profile local-gpu -d --build
    
    # or
    
    # Start Sim Studio with local model support if you don't have nvidia GPU
    docker compose up --profile local-cpu -d --build

The application will now be configured to use your local models. You can access it at http://localhost:3000/w/.

  1. Open VS Code or your favorite VS Code fork (Cursor, Windsurf, etc.)
  2. Install the Remote - Containers extension
  3. Open the project in your editor
  4. Click "Reopen in Container" when prompted
  5. The environment will automatically be set up in the sim directory
  6. Run npm run dev in the terminal or use the sim-start alias
  1. Install Dependencies
# Clone the repository
git clone https://github.com/YOUR_USERNAME/sim.git
cd sim/sim

# Install dependencies
npm install
  1. Set Up Environment
# Copy .env.example to .env
cp .env.example .env

# Configure your .env file with the required environment variables:
# - Database connection (PostgreSQL)
# - Authentication settings (Better-Auth Secret)

⚠️ Important Notes:

  • If RESEND_API_KEY is not set, verification codes for login/signup will be logged to the console.
  • You can use these logged codes for testing authentication locally.
  • For production environments, you should set up a proper email provider.
  1. Set Up Database
# Push the database schema
npx drizzle-kit push
  1. Start Development Server
# Start the development server
npm run dev
  1. Open http://localhost:3000 in your browser

We welcome contributions! Please see our Contributing Guide for details.

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.

Made with ❤️ by the Sim Studio Team

联系我们 contact @ memedata.com