terlik.js is a fast, zero-dependency, multilingual profanity detection and filtering engine built for real-world text. Designed Turkish-first, it goes far beyond simple blacklists by detecting leet speak, character repetition, separators, zero-width characters, mixed punctuation and language-specific evasion patterns. Its Turkish suffix engine can recognize thousands of derived forms from a compact set of roots, while carefully tuned whitelists help minimize false positives. It ships with Turkish, English, Spanish and German support, offers strict, balanced and loose detection modes, and can be extended with custom words or entirely new language packs. Built in TypeScript with ESM and CJS support, terlik.js runs across Node.js, Bun, Deno, browsers, Cloudflare Workers and edge runtimes, with no external dependencies, a compact bundle and sub-millisecond checks after warmup.
Loading the project