WHy not Docker install???

[gitmeatarru]

Hi.
The system is interesting, but why is there no docker file?
A lot of things need to be installed on a computer, and why? And if I don't need it tomorrow, then should I clean the system of everything that is required there???

Be more versatile!

So far, this is not an easy installation, it would be much faster with docker.

You're also going the wrong way by setting api keys through environment variables, it all has to be configured in the environment itself, you're not releasing a simple script for something there, it's a full-fledged environment, so add provider and model settings to the UI!
For example:
Add a new provider - add a model
(besides, I see a lot of complaints that you didn't do it and people ask how to add their models, I basically don't use openai, I use local models or openrouter)

[gitmeatarru]

And more...

If you specify ENV OPENAI_API_KEY, please add ENV BASE_PATH so that people can connect other providers at least this way!

[gitmeatarru]

And anyway, what do you have to do with openai???
Is the development Chinese or not???
Why are you playing along with openai if you have wonderful models like qwen???
It looks strange...

[NA-Wen]

Thanks for the feedback. Just to clarify:

1. Docker support is on our roadmap, and we plan to provide official images for easier deployment and environment isolation.
2. Configuration is already supported via the UI — you can configure the API key, model, and provider there. Environment variables such as OPENAI_API_KEY are only used as global defaults to avoid duplicating the same credentials across agents.
3. Qwen is supported. We use the OpenAI-compatible API as a unified interface. In ChatDev 2.0, the “provider” refers to the API protocol type, not the actual model vendor — so in most cases you can simply select openai, since Qwen, OpenRouter, and many local backends all expose the same OpenAI-compatible interface. This means the system is not OpenAI-only.