MCP Server

gitinspect includes a built-in Model Context Protocol (MCP) server, making it directly invokable by AI coding agents.

What is MCP?

The Model Context Protocol is an open standard that enables AI applications to securely connect to external data sources and tools. It provides a standardized way for LLMs to access and interact with external systems.

Starting the MCP Server

gitinspect mcp

This starts an MCP server over stdio — the standard transport for AI agent integration. The server listens for JSON-RPC messages on stdin and writes responses to stdout.

Available Tools

inspect_repo

Inspect a Git repository and return a structured snapshot with file contents, dependencies, and stats.

Parameters:

Parameter	Type	Required	Default	Description
repo	string	✅	—	Local path or remote URL
format	string	❌	json	Output format: json, text, yaml
max_tokens	number	❌	6000	Maximum token budget
max_files	number	❌	0	Max files (0 = unlimited)
include	string	❌	—	Include glob patterns (comma-separated)
exclude	string	❌	—	Exclude glob patterns (comma-separated)
strip	boolean	❌	false	Strip comments and blank lines
no_cache	boolean	❌	false	Disable caching

list_repo_files

List all files in a repository with priority scores and token estimates — useful for deciding which files to inspect before reading contents.

Parameters:

Parameter	Type	Required	Default	Description
repo	string	✅	—	Local path or remote URL
include	string	❌	—	Include glob patterns (comma-separated)
exclude	string	❌	—	Exclude glob patterns (comma-separated)

Two-Step Inspection Pattern

The recommended pattern for AI agents is a two-step approach:

1. List files with list_repo_files to understand the repository structure
2. Inspect with inspect_repo using targeted include/exclude patterns and appropriate max_tokens

This avoids wasting the token budget on files the agent doesn't need.