A proof of concept of a LLM-based developer sidekick that can make changes in existing code repositories. GPTinker is a Next.js application written in TypeScript that helps you navigate and modify codebases.
Short video demo: https://www.youtube.com/watch?v=XgMKCeiUDQc
- Node.js 12.x or later
- npm, yarn, or pnpm package manager
Create a .env.local file in the root directory of the project, and add your environment variables. Use the .env file as a reference. Here's an example of the required environment variables:
OPENAI_API_KEY=your_api_key_here
OPENAI_API_URL=https://api.openai.com/v1/chat/completions
OPENAI_MODEL=gpt-4
BASE_PATH=your_base_path_here
APP_DESCRIPTION="It's a Next.js application written in Typescript called GPTinker - an AI developer assistant that helps you navigate and modify codebases"
dev: Run the app in development mode.build: Build the app for production.start: Start a production server.lint: Check for linting issues.test: Run tests using Jest.countLoc: Count lines of code in the project.
To run the app in development mode first install the dependencies and then run the app:
npm install
npm run dev
# or
yarn
yarn devThe app will be available at http://localhost:3000.
To run tests:
npm run test
# or
yarn test
There is also a Dockerfile included for one-click running. Just run the command docker-compose up and the app will start and be available at http://localhost:3000.
When running the app from Docker remember that the app will not have full access to your local filesystem, so if you just want to test it out on itself put /app in the Base Path option.