Your browser. Your desktop. Your next task.
A small command-line AI assistant for Chrome and Windows apps. (Plan to develop in app)
I am building ChexsaAI to automate browser and desktop tasks at the lowest cost I can achieve. I am focusing on fewer tokens, faster execution, and reliable results with low-cost models.
I am working on benchmarks across different tasks, models, and agents to compare cost, speed, token use, and success rates.
Long term, I want ChexsaAI to work for hours without supervision.
You can watch this for setup guidelines and also for a demo: https://jumpshare.com/share/3eQ07EQeuNmfQOjPcF47
You need Windows and Python 3.11 or newer.
Open PowerShell and run:
git clone https://github.com/MashrafeeAryan/ChexsaAI.git
cd ChexsaAIRun these commands from the repository folder:
python -m venv .venv
.\.venv\Scripts\python.exe -m pip install -e .Create a .env file beside cli_.py. For Groq:
LLM_API_KEY=your-key
LLM_BASE_URL=https://api.groq.com/openai/v1
LLM_MODEL=openai/gpt-oss-120bKeep your API key private.
.envis excluded from Git.
Other OpenAI-compatible providers can use their own base URL and model. Desktop
mode requires a model with native tool calling. LLM_PROVIDER is not used.
Choose a mode. Browser and desktop run separately for now.
.\.venv\Scripts\chexsa.exe.\.venv\Scripts\chexsa.exe browserChexsa waits for your request before opening an app.
| Mode | Example request |
|---|---|
| Browser | Open YouTube and search for JavaScript tutorials. |
| Desktop | Write 300 words in Microsoft Word. |
With the virtual environment activated, use python cli_.py desktop
for desktop mode or python cli_.py browser for browser mode.
Leave the chat: type exit or quit. Cancel a request: press Ctrl+C.
Windows may require you to bring the app to the front before keyboard input continues.
| Guide | What you will find |
|---|---|
| File guide | A simple explanation of each file. |
| Architecture | How the browser and desktop pieces connect. |