dmtrkovalenko/fff

★ 5,993RustAudience · developerComplexity · 2/5Setup · easy

Mindmap

mindmap
  root((fff))
    What it does
      In-memory index
      Frecency ranking
      Fuzzy search fallback
    Surfaces
      Neovim plugin
      MCP server
      Pi agent extension
    Search Features
      Typo tolerance
      Smart-case names
      Git status labels
    AI Integration
      Claude Code
      Cursor
      Cline
    Setup
      One-line install
      Lazy.nvim plugin
      Prebuilt binary

mindmap root((fff)) What it does In-memory index Frecency ranking Fuzzy search fallback Surfaces Neovim plugin MCP server Pi agent extension Search Features Typo tolerance Smart-case names Git status labels AI Integration Claude Code Cursor Cline Setup One-line install Lazy.nvim plugin Prebuilt binary

Click or tap to explore — scroll the page freely

Things people build with this

USE CASE 1

Replace slow file finder lookups in Neovim with FFF's indexed search when working in large codebases.

USE CASE 2

Give AI coding assistants like Claude Code or Cursor a faster, context-aware file search tool via an MCP server.

USE CASE 3

Search a large codebase with automatic fuzzy fallback when an exact match returns no results.

Tech stack

RustNeovimMCP

Getting it running

Difficulty · easy Time to first run · 5min

License terms are not described in the explanation.

In plain English

FFF is a file search library and toolkit designed to be faster than general-purpose command-line tools like ripgrep or fzf when a program needs to search files repeatedly. Its speed advantage comes from keeping an in-memory index of the file system rather than scanning the disk each time. It also tracks which files you open most frequently (a concept called frecency, combining frequency and recency) and ranks those files higher in search results. The project serves several audiences. As a Neovim plugin, it replaces the default file finder and live grep with FFF's indexed search, making navigation faster in large codebases. As an MCP server, it gives AI coding tools like Claude Code, Cursor, and Cline a faster and more token-efficient search tool that understands code structure and git state. There is also an extension for the Pi agent. Each surface uses the same Rust core. Search in FFF handles typos and partial matches automatically. If an exact search returns no results, the tool retries as a fuzzy search. It also treats smart-case queries (like IsOffTheRecord) with awareness of snake_case and other naming styles. On the git side, the tool annotates modified, untracked, and staged files so AI agents can prioritize files that are actively being changed. Installing the MCP server version takes a single shell command on Linux and macOS, or a PowerShell command on Windows. The install script prints instructions for connecting the server to your chosen client. After connecting, agents can access the ffgrep, fffind, and fff-multi-grep tools. The Neovim plugin installs through standard plugin managers like lazy.nvim and downloads a prebuilt binary rather than requiring a local Rust build. The README lists configuration options and commands for each surface, with sensible defaults that work for most users without adjustment.

Copy-paste prompts

Prompt 1

I want to install the fff MCP server and connect it to Claude Code. What is the install command and how do I add it to my MCP config file?

Prompt 2

I'm using the fff Neovim plugin but the frecency ranking doesn't seem to reflect the files I actually open most. How does frecency tracking work and how do I reset it?

Prompt 3

I want fff to help my AI agent find modified and staged files first during a coding session. How does fff annotate git status in its search results?

Open on GitHub → Explain another repo

← dmtrkovalenko on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.