请问个人 pc 可以支持跑起来吗

[marianasignal]

[HeRunming]

Yes—for local development, demos, and lightweight pipeline workloads, DataFlow-WebUI can run on a personal computer. It is not a hosted-only service.

For the full WebUI profile, you will need:

• Python 3.10+ with pip
• Node.js 20+ with npm
• Sufficient memory and disk space for the Python dependencies and frontend build
• A supported agent CLI, such as Claude Code or Codex, if you want to use the browser chat panel

A local GPU is not required to start the WebUI or Harness. API-backed LLM operators can use remote model services. A CUDA-capable GPU is only needed when the operators you choose perform local model inference or other GPU-intensive processing.

On Linux or macOS, you can install and start the full WebUI with:

git clone https://github.com/OpenDCAI/DataFlow-WebUI.git
cd DataFlow-WebUI

./install.sh --profile webui --check
./install.sh --profile webui
./scripts/start.sh

Then open:

http://localhost:8000/

For a lighter installation without the visual canvas or Node.js, use the Harness profile:

./install.sh --profile harness
./scripts/start.sh

If you only need the agent skills and do not need a local server, use:

./install.sh --profile skills

On Windows, WSL2 or another Linux-compatible environment is currently the safest option because the installation and startup workflow uses Bash scripts.

In practice, the main limitation of a personal computer is usually the size of the datasets, models, and operators being executed—not the WebUI itself.