SpecLint – Spec checker for non-engineer AI builders

Help non-technical indie makers write specs AI can execute without endless rework

Problem

Designers, PMs, and operators are now building full products with AI, but they often describe requirements only by feel, causing AI to repeatedly scrap and rewrite. One maker said their coding level was roughly reading red error text, and they spent two days directing AI to build a mini-app by constantly saying the collision detection was wrong and to try again. Without structured specs, AI just guesses, burning tokens and time.

Solution

A minimalist web tool that guides users through key questions in natural language (who the user is, core flow, edge cases, technical constraints) and auto-generates a structured development spec. Templates are based on common indie projects, and the output can be pasted directly into Cursor, Claude Code, or other coding agents as a consistent reference.

Users

Non-engineering indie makers such as designers, PMs, and operators, plus beginners using AI to write code.

MVP

Pick a scenario template (WeChat mini-program, iOS utility, simple web app); ask 8-12 key questions via a wizard; generate a Markdown spec with entities, flows, and edge cases; export as a system prompt for coding agents; provide a checklist comparing against sample projects.

Revenue

Free for one spec, then pay-per-spec or monthly subscription; also a template marketplace with revenue share.

Why now

Vibe coding lowered the barrier, bringing in many non-technical builders, but AI still needs clear input to produce usable results. Spec checking is an overlooked step.