What ripgrep does well
ripgrep searches directory trees for regular-expression matches while applying repository-friendly filtering by default. It respects ignore files such as .gitignore, skips hidden files and binary files unless asked otherwise, and provides explicit glob and file-type controls. Those defaults make common codebase searches concise without preventing deeper searches when the task requires them.
Search controls and predictable escape hatches
The command supports file-type filters, include and exclude globs, context lines, replacement output, compressed-file search, multiple regex engines, and machine-readable output. When default filtering hides relevant data, --hidden, --no-ignore, --text, and the cumulative -u flag provide explicit escape hatches. The trade-off is that users must understand which filter is excluding a file before broadening the search safely.
Measured search results
We ran ripgrep 15.2.0 (official x86_64 Linux release binary, PCRE2 enabled) on 28 September 2026 against the pinned Linux 6.12 source tarball from kernel.org (86,618 files, about 1.5 GB extracted), on a shared 8-vCPU cloud VM with a warm page cache. Timed with hyperfine (3 warmup runs, 10 measured runs per query), a repository-wide literal search for EXPORT_SYMBOL_GPL returned 20,182 matching lines across 3,481 files in a mean of 208 ms. A regex search with line numbers for probe functions averaged 219 ms, a C-only (-t c) regex search averaged 165 ms, a Rust-only glob search averaged 32 ms, a case-insensitive search averaged 208 ms, and listing the 11,305 files that contain a GPL MODULE_LICENSE line averaged 205 ms. On the same literal query, GNU grep 3.11 (grep -r) returned the identical 20,182 lines in a mean of 1.15 seconds. Commands, versions, output hashes, environment details, and every per-run timing are in the public test record. These are single-machine measurements, not universal benchmarks.
Privacy and deployment
ripgrep runs locally and does not require an account, hosted service, or network request for normal searching. That makes its default privacy posture easy to reason about: searched content stays inside the environment where the command runs. Teams still need normal shell and CI controls because command output, logs, and pipelines can expose sensitive matches independently of ripgrep itself.
Limitations
Repository-aware defaults can surprise users who expect every hidden, ignored, or binary file to be searched. ripgrep is also not a universal drop-in replacement for every historical grep pipeline, especially scripts that depend on exact POSIX behavior or platform-specific flags. Finally, our measured run covers one shared Linux VM with a warm page cache. It does not measure cold-cache I/O, macOS or Windows behavior, or searches inside git checkouts, where .gitignore rules can shrink the searched file set.
Verdict
ripgrep is an excellent default search tool for developers, maintainers, and agents working across source repositories. Its strongest quality is the combination of speed, relevant default filtering, and precise opt-in controls. Learn the ignore and hidden-file rules before using it for audits that require exhaustive filesystem coverage.