Local History MCP Server

Access Cursor/VS Code Local History through the Model Context Protocol for AI-powered data recovery

Overview

This MCP server bridges the gap between AI assistants and editor Local History, enabling data recovery and enhanced context awareness. Unlike undo/redo, Local History captures file snapshots at save points, preserving work that would otherwise be lost.

Problem: Current AI assistants (Claude, Cursor AI, GitHub Copilot) cannot access Local History data despite having extensive diagnostic capabilities.

Solution: Simple MCP server providing direct access to Cursor/VS Code Local History for personal use.

Installation

Option 1: Package Manager (Recommended)

Install using your preferred package manager:

# npm
npx local-history-mcp

# pnpm
pnpm dlx local-history-mcp

# yarn
yarn global add local-history-mcp

# bun
bunx local-history-mcp

Option 2: From Source

git clone https://github.com/xxczaki/local-history-mcp.git && cd local-history-mcp
pnpm install
pnpm build
pnpm start

MCP Tools

Tool	Description
list_history_files	List all files with Local History
get_file_history	View complete history for a file
get_history_entry	Get specific history entry content
restore_from_history	Restore file to previous state (with backup)
search_history_content	Search across all history entries
get_history_stats	Overview statistics

Configuration

Cursor

You can install this MCP server in Cursor using the one-click install button:

Or manually configure by following the official Cursor MCP documentation.

Claude Code & Claude Desktop

For Claude Code (CLI), install directly:

claude mcp add local-history -- npx -y local-history-mcp

For Claude Desktop, see the official MCP documentation.

VS Code

See the official documentation.

Development

# Development mode
pnpm dev

# Run tests
pnpm test

# Linting and formatting
pnpm lint

# Launch MCP Inspector
pnpm inspector

AI disclosure

This project contains code generated by Large Language Models (LLMs), under human supervision and proofreading.

License

MIT