Baseline AIgent 🤖

An intelligent AI assistant for web developers that generates reliable, cross-browser compatible code based on the Baseline standard.

This project is a submission for the Baseline Tooling Hackathon.

Project Architecture Diagram	Baseline AIgent UI

The Problem: The "AI Trust Gap"

Modern AI assistants are powerful but often unreliable for professional web development. They are trained on a decade of outdated internet content and can generate code that:

• Uses deprecated APIs.

• Lacks cross-browser compatibility.

• Relies on obsolete techniques.

This forces developers to manually verify every line of AI-generated code, defeating the purpose of an assistant.

The Solution: An AI Grounded in Truth

Baseline AIgent solves this problem by using a Retrieval-Augmented Generation (RAG) architecture. It doesn't guess based on the entire internet; its knowledge is exclusively grounded in the official web-features dataset.

This ensures that every piece of code and every answer is:

• Reliable: Based on the official Baseline standard.

• Modern: Uses the latest, safest web technologies.

• Production-Ready: Guaranteed to be cross-browser compatible.

Features

• ✅ Smart Code Generation: Ask for a function, and get a modern, Baseline-compatible implementation.

• 🔄 Legacy Code Refactoring: Provide old code (e.g., jQuery) and get a safe, vanilla JS alternative.

• ❓ Direct Q&A: Ask about any web feature's compatibility and get an instant, accurate answer based on its Baseline status.

Tech Stack

• Data Source: Baseline (web-features dataset)

• Backend: Python, FastAPI

• AI Core: LangChain, Google Gemini API (gemini-2.5-flash)

• Knowledge Base: ChromaDB (Vector Database)

• Frontend: Vanilla HTML, CSS, JavaScript

How to Run Locally

1. Clone the repository:

   git clone https://github.com/vero-code/baseline-aigent.git
   cd baseline-aigent
   

2. Set up the Python environment:

   python -m venv venv
   source venv/bin/activate  # On Windows: .\venv\Scripts\activate
   pip install -r requirements.txt
   

3. Set up your API Key:

   • Create a .env file in the root directory.

   • Add your Google AI Studio API key: GOOGLE_API_KEY="your_api_key_here"

4. Build the Knowledge Base:

   • Download the source data: python scripts/download_data.py

   • Process the data: python scripts/data_processor.py

   • Build the vector database: python scripts/knowledge_base_builder.py

5. Run the application:

   • Start the FastAPI server: uvicorn app.main:app --reload

   • Open the frontend/index.html file in your browser.

Contributing

Contributions are welcome! As this is a hackathon project, the primary goal was to build a functional MVP. If you have ideas for improvements or new features, feel free to fork the repository or open an issue to discuss them.

License

This project is licensed under the MIT License - see the LICENSE file for details.