YT · YOUTUBE// 3h agoOPENSOURCE RELEASE

FFF makes file search stateful for agents

FFF is a Rust-based file search toolkit built for AI agents and Neovim. Instead of spawning a fresh `rg` or `fzf` process every time, it keeps an index and file cache warm in memory, then layers on fuzzy path search, content grep, frecency ranking, git-aware results, and bindings for Rust, C, Node/Bun, Neovim, and MCP. The pitch is simple: if a tool is going to search the same repository repeatedly, it should behave like a long-lived service, not a disposable command.

// ANALYSIS

Hot take: this is the right shape of tool for agentic coding, because the bottleneck is not raw search speed once, it is repeated search latency across an entire session.

–Strong fit for AI agents and editors that ask many related questions over the same repo.
–The built-in memory, background indexing, and warm caches are the real differentiators, not just the fuzzy matcher.
–Frecency and git-status awareness make the ranking more useful than plain filename matching.
–It is not the right default for a single terminal grep; `rg` still wins on simplicity and zero setup.
–The tradeoff is deliberate: more memory and more moving parts in exchange for much faster repeated searches.

// TAGS

file-searchagentneovimrustfuzzy-searchgrepdevtoolopensource

DISCOVERED

3h ago

2026-04-29

PUBLISHED

3h ago

2026-04-29

RELEVANCE

8/ 10

AUTHOR

Syntax