This repository builds a server for the tabbyML clients (VSCode, Vim, IntelliJ). The code is written in Python and uses FastAPI for HTTP request handling. It utilizes the Gemini API to fetch code snippets.
The code requires the credentials of your google cloud service account, below is the official doc to generate a key. Create google account credentials
To run this project, follow these steps:
-
Create a Python Virtual Environment:
python -m venv .venv
This command creates a Python virtual environment in the
.venvfolder. -
Install Dependencies:
pip install -r requirements.txt
This command installs the required dependencies for the project.
-
Run the Server:
uvicorn server:app
This command starts the server using Uvicorn.
If you would like to contribute to this project, please follow the contribution guidelines.
The latency of the gemini api is slow so configure the tabbyML extension to incerase the completion request timeout. Below is the command to open the config file of tabby agent. Uncomment the Completion section and change timeout from 4000ms to 15000ms. Reload the extension after changes in file.
nano ~/.tabby-client/agent/config.toml