The problem
Language models can explain code, but their free-form answers are hard to use inside other tools: the format changes from one response to the next, and they sometimes return invalid output.
What we built
- Prompt engineering that asks the Gemini model for a fixed, structured answer: explanation, bugs, complexity and suggestions.
- Response validation so the API always returns reliable JSON that other software can depend on.
- A Node.js and Express REST API secured with API-key authentication and rate limiting, with a live web demo.
Built with
Node.jsExpressGemini APIREST