Delta MCP Server

Delta MCP Server

Brings git-delta's beautiful syntax highlighting and diff formatting to AI assistants, providing colorized diffs, side-by-side comparisons, blame information, and grep results with clickable browser links.

Category
Visit Server

README

Delta MCP Server

A Model Context Protocol server that brings git-delta's beautiful syntax highlighting and diff formatting to AI assistants like Cursor, Claude Code, Claude Desktop, and Warp. Instead of plain text diffs, you get clean terminal-style colorized output that looks exactly like native delta, with the added benefit of clickable browser links for better viewing.

What This Does

When you ask an AI assistant to show you a git diff or explain code changes, this server formats the output using git-delta. That means you get proper syntax highlighting, side-by-side comparisons, line numbers, and all the visual enhancements that make diffs readable. The server integrates seamlessly with MCP-compatible clients, so you can use natural language to request formatted diffs, blame information, grep results, and more.

The server also supports terminal hyperlinks, so when you view a diff in your terminal, you can click a link to open it in your browser for a better viewing experience. All formatted output is saved as HTML files that persist on disk, making it easy to share or bookmark specific diffs.

Requirements

You'll need Python 3.10 or higher, and git-delta must be installed. The easiest way to install delta is with cargo: cargo install git-delta. You'll also need Git installed, and I recommend using uv for dependency management, though pip works fine too.

Installation

There are a few ways to set this up depending on how you want to use it. For Claude Desktop users, there's an MCPB package that provides one-click installation. If you prefer pip, you can install it as a package. For development or custom setups, you can clone the repository and set it up manually.

For development setup, start by installing uv if you don't have it already. Then clone the repository, create a virtual environment with uv venv, activate it, and install the dependencies with uv pip install -e .. The server will be ready to run.

Configuration

Delta uses Git's configuration format, so you configure it by adding a [delta] section to your ~/.gitconfig file. You can set the syntax theme, enable side-by-side view, turn on line numbers, and configure other options. There's an example configuration file in the config directory that shows all the available options.

For MCP client configuration, you'll need to add the server to your client's configuration file. The exact location depends on which client you're using. For Claude Desktop, it's in the Application Support directory. For Claude Code, use the MCP settings in the application. For Cursor, you can use the UI-based setup in Settings, or manually edit the configuration file. Warp uses a similar JSON configuration format. Example configuration files are provided in the examples directory for each client.

The key thing to remember is to use absolute paths, not relative ones. On Windows, make sure to use double backslashes or forward slashes in your paths. After making configuration changes, you need to fully quit and restart your client, not just close the window.

Usage

Once configured, you can use natural language to interact with the server through your AI assistant. Ask it to show you the formatted diff for a specific commit, format the git log with patches, search for patterns in your codebase, or compare two files. The server handles all the formatting automatically.

Quick Examples for Claude

Basic usage:

  • "Show me the diff for the last commit"
  • "Format the git diff for server.py"
  • "What changed in HEAD~1?"

CLI mode (no browser links):

  • "Show me the diff in CLI mode"
  • "Use CLI mode to show the last commit"

Browser mode (with clickable links):

  • "Show me the diff and open it in browser"
  • "Format the diff with browser links"

See USAGE.md for detailed examples and all available tools.

The server provides several tools that your AI assistant can use. These include formatting git diffs, showing commit details, displaying git log with patches, showing blame information, formatting grep results, displaying merge conflicts, syntax highlighting code snippets, and comparing files directly.

There are also resources available that provide direct access to formatted data. You can access formatted diffs for specific commits, blame information for files, grep results for patterns, your current delta configuration, and a list of available syntax themes.

Security

The server includes input validation to prevent common security issues. File paths are validated to ensure they're within the git repository or current working directory, preventing path traversal attacks. Commit ranges are validated using regex patterns. Grep tools are restricted to a whitelist of safe tools: grep, ripgrep, and git-grep.

Troubleshooting

If you see an error about Delta not being found, make sure you've installed it with cargo install git-delta. You can verify the installation by running delta --version. If you have delta installed in a non-standard location, you can configure the path in the MCPB manifest.

If your Git configuration isn't loading, check that your ~/.gitconfig file exists and has the correct syntax. Delta uses INI format, not TOML. You can also set the GIT_CONFIG_GLOBAL environment variable to point to a different config file.

When the server doesn't show up in your client, the most common issues are invalid JSON in the configuration file, using relative paths instead of absolute ones, or not fully restarting the client after making changes. Make sure to completely quit the application, not just close the window.

If tool calls are failing, check the client's log files. For Claude Desktop, logs are in the Logs directory. You can also test the server manually by running uv run server.py to see if there are any errors.

Development

To run the test suite, install the development dependencies with uv pip install -e ".[dev]" and then run pytest. The tests cover validation, configuration management, and git utilities.

To build an MCPB package for distribution, install the MCPB CLI as a dev dependency with npm, then run npx @anthropic-ai/mcpb pack to create the bundle file.

For building a pip package, use python -m build to create the distribution files, then upload them with twine upload dist/*.

License

This project is licensed under the MIT License.

Acknowledgments

This server is built on top of git-delta, an excellent diff tool that provides the syntax highlighting and formatting capabilities. It uses the Model Context Protocol specification for integration with AI assistants, and the FastMCP framework for rapid development.

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
E2B

E2B

Using MCP to run code via e2b.

Official
Featured