Show HN: Prompt optimizer for vibe-coding with LLMs

I’ve been working on a small tool aimed at reducing prompt friction in vibe-coding workflows.

In practice, a lot of iteration comes from underspecified prompts: missing constraints, unclear scope, implicit assumptions, or mixed intent. This tool takes a rough, natural-language description of what you want to build and rewrites it into a more explicit, structured prompt with clearer requirements and context before it’s sent to the model.

The focus is on:

Making intent, constraints, and assumptions explicit

Reducing prompt churn and micro-iterations

Improving first-pass output quality, especially for non-technical builders

It’s primarily designed around vibe-coding use cases (rapid prototyping, AI-assisted building) and works best with Lovable/Claude-style workflows, though it’s model-agnostic in concept.

Very interested in technical feedback:

Is prompt normalization / restructuring something you’ve found valuable?

Do you solve this via system prompts, fine-tuning, or runtime prompt transforms?

Where does this break down for more complex or long-context tasks?

Happy to hear critical takes.