FFF (Fast File Finder) is a file search toolkit built for humans and AI agents. It delivers typo-resistant path and content search that feels instant even in huge repositories, thanks to a background watcher and a lightweight in-memory index. The project has collected about 11,200 stars on GitHub and is under active development in 2026.
Where tools like ripgrep and fzf rescan the filesystem on every query, FFF keeps a frecency-ranked index in memory: files you open often float to the top, partial and misspelled queries still land on the right file, and content search runs against the index rather than cold disk reads. It ships as a core engine written in Rust with official SDKs for Rust, C, Python, Bun and Node.js, plus a Lua binding that powers a popular Neovim plugin. Coding agents use it to locate relevant files in milliseconds instead of burning tokens on repeated greps, and benchmarks published by the project show it outperforming classic CLI searchers on large codebases. The toolkit also ships a terminal UI and a command-line frontend for everyday use, so it works as a faster everyday fuzzy finder even outside agent workflows. Its maintainer uses it as a production dependency of his own editor setup, which shows in the attention paid to indexing speed, memory footprint and ranking quality on repositories with hundreds of thousands of files.
FFF is completely free and open source under the MIT license, including for commercial use in your own tools and agent products.
FFF is for agent builders, editor plugin authors and developers working in very large repositories who need file and content search that returns ranked results in milliseconds.