
gitreverse
Reverse engineer any repo into it's original prompt
The Lens
By Erik Loyd, SaaS CEO and former COO/CFO of an AWS Premier Partner.
Updated Aug 2026
GitReverse takes a public GitHub repo and works backwards to the prompt that could have built it. It reads the structure, the README, and the metadata, then has an LLM write the kind of plain-English instruction you would hand Cursor or Claude Code to recreate the project. It now does the same trick on websites. A learning tool and a decent way to see how a finished codebase reads as a specification.
Running it yourself is more work than the demo suggests. It is a Next.js 16 app you clone and configure: copy the env template, supply at least one LLM provider key from Grok, OpenRouter, Azure OpenAI, or Google AI Studio, then pnpm install and pnpm dev. A GitHub token, Supabase, and Stripe are optional depending on the features you want.
The generated prompts are a starting point, not a blueprint. Reading structure and a README tells you what a project looks like from outside, which is not the same as knowing why it was built that way. Solo developers and people learning how projects are shaped get the most out of it. Teams will find it interesting rather than useful.
The catch is the license, or rather the absence of one. This repository ships no LICENSE file, which under default copyright means all rights reserved: no permission to use, modify, or redistribute, whatever public availability suggests. Running the hosted demo is fine. Forking it is not, until the author adds a license.
Free vs Self-Hosted vs Paid
source availableWhat's Free
The hosted version is free to use, and the source is publicly readable on GitHub. That is the honest extent of it.
The License Problem
There is no LICENSE file in the repository. Under default copyright law that means all rights reserved. Public visibility is not a grant of permission, so self-hosting, forking, or building on this code is not legally covered until the author publishes terms. This is almost certainly an oversight rather than a stance, and it is worth an issue rather than an assumption.
Self-Hosting Cost
Node and pnpm, plus at least one LLM provider key. Costs land on the provider you choose:
- Google AI Studio: has a free tier that covers casual use
- OpenRouter: pay per token, with some free models available
- Grok / Azure OpenAI: paid, metered per token Optional Supabase (free tier available) for history and library, and Stripe only if you are running it as a paid service yourself.
The Real Cost
Each reverse-engineering run reads repository content and generates a long prompt, so cost tracks repo size. Casual use fits inside Google AI Studio's free tier comfortably; batch-processing many repositories does not.
Free to use and free to read, but the repository ships no license at all, so self-hosting or forking it is not legally covered until the author adds one.
What to do by team size
- Solo
- free to use hosted; do not fork it until a license exists
- Small team
- free to use hosted; the missing license blocks internal deployment
- Medium team
- Skip self-hosting until licensing is resolved
- Large team
- Skip; no license means no defensible internal use
Get tools like this every Wednesday
One featured tool, three on the radar. No fluff.
A low score is not a verdict on quality. Young and niche tools start low by design. How we calculate scores
About
- Owner
- Fili (User)
- Stars
- 1,907
- Forks
- 316
Explore Further
More tools in the directory
ponytail
Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.
135.1k ★caveman
🪨 why use many token when few token do trick. Claude Code skill that cuts 65% of tokens by talking like caveman
104.8k ★cherry-studio
AI productivity studio with smart chat, autonomous agents, and 300+ assistants. Unified access to frontier LLMs
51.7k ★