smkttl/deepseek-api ? reverse-engineered prompt

Reverse engineered prompt

Build me a small Python app that lets me use DeepSeek without opening the website every time.

I want two ways to use it, a simple command line chat mode and a local server that looks like an OpenAI style API, so I can plug it into other tools easily. It should take my DeepSeek session cookie and bearer token from a tokens file or environment variables, then send messages and return the model reply in a clean way. Please support both normal chat and the reasoning mode, and make the output readable with Markdown if possible.

It would be great if I can choose between a few model names like DeepSeek V3 and R1, and have streaming output too. If anything about the current DeepSeek web interface has changed, look up the latest docs online if you need to and make it work with the newest flow. Keep it lightweight and easy to run locally.

Are you gonna build this?

make sure you review the code using coderabbit

Try freeSponsored — opens CodeRabbit in a new tab