feat: Hello AI AVS with a minimal interface

[sotoJ24]

🧠 Add Inference Step Using LLM API in AVS Handler

✨ Summary

This PR enhances the AVS by including an inference step during the request validation process using a language model API (Groq/OpenAI-compatible).

🔍 What was done

• Added environment variable support for GROQ_API_KEY
• Integrated an LLM call inside the AVS handler (handler.ts)
• The handler now performs a basic inference based on the user input
• Fallback to a mock or placeholder if no key is provided

📚 Motivation

This aligns with the AVS requirement to:

"perform some inference: either locally via small local LLM like Gemma or remotely via calls to an LLM API"

Result

🧪 How to test

Run:

npx tsx scripts/avs/index.ts