reltadev/github-assistant: Explore GitHub repositories with natural language questions

github-assistant is a PoC built using Relta and assistant-ui. You can try it out here and read more about the project in this blog.

The Relta sub-module is currently not open source. You can see the semantic layer that was generated by the library and is used in the text-to-SQL pipeline in the server-poc/semantic_layer folder. The full library is available upon request for both commercial and non-commercial uses. Send an email to amir [at] relta.dev for access.

• Python 3.9+
• npm or other Node.js package manager
• Git

1. Reach out to get access to the Relta submodule.

2. Initialize the relta submodule

git remote add template https://github.com/reltadev/poc-template.git && git submodule update --init --recursive

2. Create a virtual environment for Relta

python -m venv .venv && source .venv/bin/activate && pip install -r requirements.txt

3. Setup the .env files from .env.example

cp client-poc/.env.example client-poc/.env && cp server-poc/.env.example server-poc/.env

4. Set the following environment variables in server-poc/.env:

   • OPENAI_API_KEY: Your OpenAI API key
   • GITHUB_DATABASE_CONNECTION_URI the PG database where the GitHub data will be piped into

5. Launch the backend

cd server-poc
uvicorn server_poc.server:app --host 0.0.0.0 --port 80 --reload4

5. Launch the front-end

cd client-poc
npm install
npm run dev

We will generally give instructions on how to update Relta or the POC to handle any bugs or new features.

We would love to get new contributors! If you are interested, please reach out to amir [at] relta.dev or simon.farshid [at] outlook.com