peteromallet/desloppify
A framework that equips AI coding agents with LLM-driven tools to measure, plan, and systematically improve code quality.

Desloppify is an agent harness that provides AI coding agents with both mechanical detection (dead code, duplication, complexity metrics) and subjective LLM review (naming, abstractions, module boundaries). It combines these into a single quality score that agents can work toward over multiple sessions, with state persisting across scans to incrementally improve the codebase. Supports 29 languages with deep plugin support for TypeScript, Python, C#, C++, Dart, Go, and Rust, and generates scorecard badges for GitHub profiles.