// the find
BurntSushi/ripgrep
ripgrep recursively searches directories for a regex pattern while respecting your gitignore
ripgrep is a recursive code search tool that's faster than grep, ag, and git-grep on nearly every benchmark that matters. It respects .gitignore by default, handles Unicode correctly without a performance penalty, and ships static binaries for every major platform. The target user is any developer who searches codebases regularly and is tired of waiting.
The regex engine is genuinely fast — built on Rust's finite automaton implementation with SIMD literal optimizations, not just wrapped PCRE. The automatic filtering (gitignore, hidden files, binaries) is on by default and composable: -uuu turns it all off, -tc limits to C files, and you can define custom file types. The workspace is split into individually publishable crates (ignore, globset, grep, searcher, printer), so the gitignore-walking and search logic are usable as libraries without pulling in the CLI. PCRE2 is available as an opt-in flag for look-around and backreferences when the default engine isn't enough.
Not POSIX-compliant, so it can't replace grep in scripts that need to run in arbitrary environments — the README is honest about this but it will still catch people. Performance degrades badly on patterns without literal prefixes (the [A-Za-z]{30} case takes 15 seconds), which is the same cliff GNU grep hits but worth knowing before you assume rg is always fast. The --json output is the only structured format, so if you're building tooling that needs to process results programmatically you're parsing JSON or writing your own sink against the library API. Windows support is real but second-class: some edge cases in gitignore parsing and file encoding detection are documented as known gaps.