MCP 服务器

git-slim

A token-optimized Git MCP server that reduces context window tokens by 59% while preserving full functionality, enabling AI assistants to interact with Git repositories using only 5 grouped tool operations.

README

git-slim

Git MCP server optimized for AI assistants — Reduce context window tokens by 59.0% while keeping full functionality. Compatible with Claude, ChatGPT, Gemini, Cursor, and all MCP clients.

What is git-slim?

A token-optimized version of the Git Model Context Protocol (MCP) server.

The Problem

MCP tool schemas consume significant context window tokens. When AI assistants like Claude or ChatGPT load MCP tools, each tool definition takes up valuable context space.

The original mcp-server-git loads 12 tools consuming approximately ~7,926 tokens — that's space you could use for actual conversation.

The Solution

git-slim intelligently groups 12 tools into 5 semantic operations, reducing token usage by 59.0% — with zero functionality loss.

Your AI assistant sees fewer, smarter tools. Every original capability remains available.

Performance

Metric	Original	Slim	Reduction
Tools	12	5	-58%
Schema Tokens	1,086	400	63.2%
Claude Code (est.)	~7,926	~3,250	~59.0%

Benchmark Info

Original: mcp-server-git@latest

Schema tokens measured with tiktoken (cl100k_base)

Claude Code estimate includes ~570 tokens/tool overhead

Quick Start

One-Command Setup (Recommended)

# Claude Desktop - auto-configure
npx git-slim --setup claude

# Cursor - auto-configure
npx git-slim --setup cursor

# Interactive mode (choose your client)
npx git-slim --setup

Done! Restart your app to use git.

CLI Tools (already have CLI?)

# Claude Code (creates .mcp.json in project root)
claude mcp add git -s project -- npx -y git-slim@latest

# Windows: use cmd /c wrapper
claude mcp add git -s project -- cmd /c npx -y git-slim@latest

# VS Code (Copilot, Cline, Roo Code)
code --add-mcp '{"name":"git","command":"npx","args":["-y","git-slim@latest"]}'

Manual Setup

<details> <summary>Click to expand manual configuration options</summary>

Claude Desktop

Add to your claude_desktop_config.json:

OS	Path
Windows	`%APPDATA%\Claude\claude_desktop_config.json`
macOS	`~/Library/Application Support/Claude/claude_desktop_config.json`

{
  "mcpServers": {
    "git": {
      "command": "npx",
      "args": ["-y", "git-slim@latest"]
    }
  }
}

Cursor

Add to .cursor/mcp.json (global) or <project>/.cursor/mcp.json (project):

{
  "mcpServers": {
    "git": {
      "command": "npx",
      "args": ["-y", "git-slim@latest"]
    }
  }
}

</details>

How It Works

MCPSlim acts as a transparent bridge between AI models and the original MCP server:

┌─────────────────────────────────────────────────────────────────┐
│  Without MCPSlim                                                │
│                                                                 │
│  [AI Model] ──── reads 12 tool schemas ────→ [Original MCP]    │
│             (~7,926 tokens loaded into context)                 │
├─────────────────────────────────────────────────────────────────┤
│  With MCPSlim                                                   │
│                                                                 │
│  [AI Model] ───→ [MCPSlim Bridge] ───→ [Original MCP]           │
│       │                │                      │                 │
│   Sees 5 grouped      Translates to        Executes actual   │
│   tools only         original call       tool & returns    │
│   (~3,250 tokens)                                              │
└─────────────────────────────────────────────────────────────────┘

How Translation Works

AI reads slim schema — Only 5 grouped tools instead of 12
AI calls grouped tool — e.g., interaction({ action: "click", ... })
MCPSlim translates — Converts to original: browser_click({ ... })
Original MCP executes — Real server processes the request
Response returned — Result passes back unchanged

Zero functionality loss. 59.0% token savings.

Available Tool Groups

Group	Actions
`mutation`	4
`navigation`	3
`query`	3

Plus 2 passthrough tools — tools that don't group well are kept as-is with optimized descriptions.

Compatibility

✅ Full functionality — All original mcp-server-git features preserved
✅ All AI assistants — Works with Claude, ChatGPT, Gemini, Copilot, and any MCP client
✅ Drop-in replacement — Same capabilities, just use grouped action names
✅ Tested — Schema compatibility verified via automated tests

FAQ

Does this reduce functionality?

No. Every original tool is accessible. Tools are grouped semantically (e.g., click, hover, drag → interaction), but all actions remain available via the action parameter.

Why do AI assistants need token optimization?

AI models have limited context windows. MCP tool schemas consume tokens that could be used for conversation, code, or documents. Reducing tool schema size means more room for actual work.