mcp server

fastkb

Minimalist CLI to index & search local files instantly. Powered by SQLite FTS5 + BM25. Zero dependencies, offline, in-memory mode.

egor-morozik/fastkbcommunitylistedunreviewed

Use tool

Set up

Runs on your computer. Pick your agent and operating system.

Before you start

uv (`curl -LsSf https://astral.sh/uv/install.sh | sh`)

~/Library/Application Support/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "fastkb": {
      "command": "uvx",
      "args": [
        "fastkb"
      ]
    }
  }
}

same format Cursor, GitHub Copilot CLI, Junie, JetBrains AI Assistant, Augment (Auggie CLI), Factory Droid, Kiro, Devin CLI +12

Values in angle brackets are placeholders. Restart Claude Desktop after editing its config.Claude Desktop docs

README

synced Oct 26, 2026

FastKB

A minimalist CLI tool for indexing and searching local text files using SQLite FTS5. Fast, offline, and zero-dependency.

Features

Fast Indexing: Recursively scans directories and files.
Full-Text Search: Instant ranked search (BM25) via SQLite FTS5.
Auto-Sync: SQL triggers keep the search index updated automatically.
In-Memory Mode: Run entirely in RAM for zero-disk-footprint operations.

Installation

Core (CLI + FTS5 search):

pip install fastkb

With AI/MCP support:

pip install "fastkb[mcp]"

Quick Start

1. Initialize

Create the database and search triggers:

fastkb init

2. Index

Scan folders or specific files:

fastkb index ./docs ./notes.txt

3. Search

Search for content (returns top 5 results by default):

fastkb query "your search term"

Advanced Usage

In-Memory Mode

Use --memory (-m) to run the database entirely in RAM. Ideal for CI/CD pipelines, automated tests, or temporary scans that shouldn't leave files on disk.

fastkb --memory index ./src
fastkb --memory query "timeout error"

Note: The in-memory database is transient and clears when the process exits. Best used for single-session workflows or scripted runs.

CLI Reference

Commands

Command	Arguments	Description
`init`	-	Initialize database and FTS index
`index`	`paths [paths ...]`	Index directories or files recursively
`query`	`text`	Search the indexed knowledge base

Global & Query Options

Flag	Scope	Default	Description
`-m, --memory`	Global	`false`	Use in-memory database (non-persistent)
`-l, --limit`	`query`	`5`	Maximum number of search results to return

AI Integration

FastKB supports the Model Context Protocol (MCP) for seamless integration with Claude Desktop, Cursor, Cline, Continue, and other AI assistants.