Hi Product Hunt 👋 I built bonsai-lint to make human collegues and AI-agents write higher quality code across projects/monorepos. Cognitive complexity measures how hard code is to read, not how hard it is to test. A switch with twenty arms is fine; three nested ifs are not. bonsai-lint ranks every function by that score, worst first, and fails CI above a threshold (15 by default). What makes it different: • One binary for every language. No runtime, no plugins, nothing to configure. A mixed codebase is scored in one pass on one metric, and the same logic gets the same number in every language. That's tested, so one threshold means the same thing across a whole monorepo. • Fast. 1.26 million lines in three languages, in one pass, in 0.77 seconds. • It never runs your code. It reads syntax only, so untrusted source is safe to scan. • Callbacks compound. A closure inside a loop inside a condition is scored at its real depth, so a callback pyramid shows up as one hard function, not several easy ones. • It scores code outside functions too. Templates, scripts, route files and module-level setup are scored. On legacy code, that's often where the worst of it hides. • You can adopt it on day one. Record today's findings as a baseline and CI fails only on regressions. With --strict-baseline, the baseline has to tighten as the code improves. Try it without installing anything: npx bonsai-lint src/ uvx bonsai-lint src/ or paste code into the playground: https://bonsai.kauneckas.dev/play/ It's also on Homebrew, Composer, go install and cargo. There's a VS Code extension that scores your code as you type, a pre-commit hook, and pull request annotations with --format github. Two questions for you: which language should it read next, and what threshold would you set for your team?

bonsai-lint

Multi-language cognitive complexity linter, in one binary

Multi-language cognitive complexity linter, in one binary