This is a .NET linter that catches the shortcuts LLMs take to make tests pass without fixing the actual problem. It detects disabled tests, suppressed warnings, empty catch blocks, arbitrary delays, and project-level warning suppressions. You run it after every code change, ideally as a post-edit hook, and it compares against a baseline so you're only catching new slop rather than inherited tech debt. The strict mode treats everything as an error, which is the right call when working with AI coding assistants that will happily disable a flaky test instead of fixing the race condition. It's opinionated tooling for a real problem: reward hacking in generated code.
npx skills add https://github.com/aaronontheweb/dotnet-skills --skill dotnet-slopwatch