A minimal agentic AI demo that analyzes code to infer intent, assumptions, and risks — without rewriting or modifying the code.
This is a small experiment to explore agentic AI patterns in a minimal way.
Given a code snippet, the agent:
- Infers the intent of the code
- Highlights potential risks and assumptions
The goal is to understand how structured reasoning can make AI outputs more useful and reliable.
- Accepts a code snippet
- Infers the developer’s intent
- Identifies assumptions, risks, and edge cases
- Provides structured feedback
- Does not rewrite code or suggest features
- Node.js 18+
- An OpenAI API key
git clone https://github.com/Cybernaut365/intent-aware-coding-agent
cd intent-aware-coding-agent/backendnpm installCreate a .env file in the backend folder:
OPENAI_API_KEY=your_openai_api_key_hereReplace
your_openai_api_key_herewith your actual OpenAI key. Keep this file private.
npm run devOpen frontend/index.html in your browser. You’ll see:
- A textarea to paste your code
- A button: Analyze Code
- An output section for the AI’s analysis