an agent harness to make your codebase beautiful
Give your AI coding agent a north star. Desloppify scans, scores, and systematically improves code quality — mechanical issues and subjective ones — across 29 languages. The score resists gaming. The only way up is to actually make the code better.
Desloppify combines mechanical detection (dead code, duplication, complexity) with subjective LLM review (naming, abstractions, module boundaries). It scans your codebase, builds a prioritized plan, then works through a fix loop — one issue at a time. State persists across sessions so it chips away over multiple runs, and the scoring is designed to resist gaming. A score above 98 should correlate with a codebase a seasoned engineer would call beautiful.
Vibe coding gets things built fast. But the codebases it produces rot in ways that are hard to see — abstractions drift, naming degrades, error handling diverges. It works, but working in it gets worse.
LLMs are good at spotting this now, if you ask the right questions. The bet: an agent with the right framework can hold a codebase to the standard that used to require a senior engineer paying close attention over months.
Join the community of vibe engineers who want to build beautiful things → Discord
Pulled live from GitHub.
The $Desloppify Experiment
Community crypto folks created a Solana token for Desloppify and gave us ~7% of the supply plus creator fees. We're putting 100% of it toward boosting the open source ecosystem — bounties, challenges, and funding.
Our first initiative: a $1,000 bounty to find something poorly engineered in our 91k-line AI-built codebase. 262 comments. Winner: @agustif. Now live: $1,000 if Desloppify does something stupid when refactoring your codebase.