HG MCP Server

HG MCP Server

Enables AI assistants to interact with Mercurial repositories, supporting version control operations like cloning, committing, branching, merging, and remote synchronization through natural language.

Category
Visit Server

README

HG MCP Server

PyPI Version MCP Badge

A Model Context Protocol (MCP) server for Mercurial repository interaction, written in Python 3.10+ with AsyncIO.

Features

Core Version Control

  • Repository Creation: Create new repositories (hg_init) or clone existing ones (hg_clone)
  • Status & Diff: View working directory status (hg_status) and line-by-line changes (hg_diff)
  • Commit Management: Add files (hg_add), commit changes (hg_commit), remove files (hg_remove), and amend the last commit (hg_amend)
  • Stashing: Temporarily shelve changes (hg_shelve) and restore them (hg_unshelve)
  • File Inspection: View file content at any revision (hg_cat) and rename or move files (hg_rename)

Branching & Navigation

  • Navigation: Update to any revision, branch, or bookmark (hg_update)
  • Named Branches: Create and list permanent Mercurial branches (hg_branch)
  • Bookmarks: Manage lightweight, Git-like pointers (hg_bookmark, hg_bookmarks)
  • Advanced Topics: WIP feature isolation with lightweight branches (hg_topic, hg_topics, hg_topic_current)
  • Phases: Manage changeset phases for safe history rewriting (hg_phases)
  • Tags: Create, remove, and list repository tags (hg_tag, hg_tags)

Remote Synchronization

  • Push/Pull: Sync changes with remote repositories (hg_push, hg_pull)
  • Remote Insights: Preview incoming or outgoing changesets (hg_incoming, hg_outgoing)
  • Path Management: List and manage configured remote paths (hg_paths)
  • Branch Heads: Identify branch heads across the repository (hg_heads)

History Rewriting & Recovery

  • Rebase & Strip: Move changesets (hg_rebase) or permanently remove them (hg_strip)
  • Evolve Toolset: Auto-amend (hg_absorb), fold changesets (hg_fold), split changesets (hg_split), uncommit (hg_uncommit), prune (hg_prune), rewind for undo (hg_rewind), edit metadata (hg_metaedit)
  • Stack Navigation: Move through topic stacks (hg_next, hg_previous, hg_stack)
  • Interactive Editing: Modify a linear series of changesets non-interactively (hg_histedit)
  • Graft & Transplant: Copy changesets via merge machinery (hg_graft) or patch-based (hg_transplant)
  • Backout: Reverse the effects of earlier changesets (hg_backout)
  • Evolve: Track the evolution of rewritten changesets (hg_evolve)

Merge & Conflict Resolution

  • Merging: Combine changes from different branches (hg_merge)
  • Conflict Management: List and track files with unresolved merge conflicts (hg_resolve)
  • Revert: Discard uncommitted changes and restore files to their last committed state (hg_revert)

Repository Inspection & Diagnostics

  • Blame/Annotate: Show line-by-line changeset information (hg_annotate)
  • Bisect: Binary search for regression-introducing changesets (hg_bisect)
  • File Listing: List all tracked files in the current revision (hg_files)
  • State Summary: Get a concise summary of the working directory state (hg_summary)
  • Integrity & Identity: Verify repository integrity (hg_verify) and identify the current revision (hg_identify)
  • Patch Management: Export changesets as patches (hg_export) or import patch files (hg_import)

Technical Features

  • Async I/O: Asynchronous command execution for high-performance operations
  • Git Integration (hg-git): Automatic detection of Git-backed repos and branch mapping (hg_git)
  • HTTP Transports: Streamable HTTP and SSE support for web-based clients (e.g., llama.cpp WebUI)
  • Security: Jail path restriction (required for HTTP) and secure API key authentication
  • JSON Output: Automatic JSON formatting for 20+ commands for easy machine parsing
  • Large File Support: Management of binaries outside normal history (hg_largefiles)
  • Diagnostics & Help: List extensions (hg_extensions), config (hg_config), and access built-in help (hg_help)
  • Reliability: 349 tests with 87% coverage, using tmpfs on Linux for extremely fast verification
  • 62 MCP Tools: Comprehensive coverage of Mercurial 7.1.x standard commands and built-in extensions

Installation

# Clone the repository
git clone <repository-url>
cd hg-mcp

# Create and activate virtual environment
python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

# Install in editable mode
pip install -e .

# Optional: Install with performance enhancements
pip install uvloop  # On Unix/macOS
pip install winloop  # On Windows

Usage

Standard Input/Output (stdio) - Default Mode

# Activate your virtual environment first
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
hg-mcp

This starts the MCP server using stdio transport, which can be used with MCP clients like Claude for Desktop, Qwen-Coder, Gemini-CLI, and OpenCode.

HTTP Transport (SSE and/or Streamable HTTP)

For web-based MCP clients like llama.cpp WebUI, you can run the server with HTTP transport:

# SSE only (compatible with llama.cpp WebUI)
hg-mcp --transport sse --port 8000

# Streamable HTTP only
hg-mcp --transport streamable-http --port 8000

# Both SSE and Streamable HTTP on the same server
hg-mcp --transport sse streamable-http --port 8000

Endpoints:

  • SSE: http://localhost:8000/sse
  • Streamable HTTP: http://localhost:8000/mcp

Options:

  • --transport: One or more transport protocols (stdio, sse, streamable-http)
    • Default: stdio
    • Multiple values: --transport sse streamable-http
  • --port: Port to listen on (default: 8000)
  • --host: Host to bind to (default: 0.0.0.0)
  • --api-key: Optional. Enable mandatory API key authentication. When set, clients must provide this key in their request headers.
  • --jail: Required for HTTP transports. Restrict repository access to this directory tree for security.
  • --mercurial: Optional. Path to the Mercurial (hg) executable. Defaults to hg (resolved from PATH). Example: --mercurial /usr/local/bin/hg

Security Features

When exposing the MCP server over TCP/IP, two security mechanisms are available:

1. Jail Path Restriction (--jail)

Required for all HTTP transports. This restricts repository access to a specific directory tree, preventing unauthorized access to paths outside the jail:

# Restrict access to /home/user/projects and its subdirectories
hg-mcp --transport streamable-http --port 8000 --jail /home/user/projects

2. API Key Authentication (--api-key)

Optional. You can require clients to authenticate by providing a valid API key. When this option is enabled, the server will reject any request that does not include a matching X-API-Key or API-Key header:

# Enable mandatory API key authentication
hg-mcp --transport streamable-http --port 8000 --jail /home/user/projects --api-key your-secret-key

Integration with AI Assistants

To use this MCP server with your AI coding assistant, point to the correct executable path in your virtual environment.

Configuration Examples

Claude Desktop / Qwen-Coder / Gemini-CLI

{
  "mcpServers": {
    "hg-mcp": {
      "command": "/path/to/your/venv/bin/hg-mcp"
    }
  }
}

OpenCode

{
  "mcp": {
    "hg-mcp": {
      "type": "local",
      "command": ["/path/to/your/venv/bin/hg-mcp"],
      "enabled": true
    }
  }
}

Requirements

  • Python 3.10+
  • Mercurial installed and available in PATH
  • MCP-compatible client

Development

Running Tests

# Install development dependencies
pip install -e ".[dev]"

# Run all tests with coverage
./scripts/runtest.sh

Code Quality

# Run linting
./scripts/lint-check-and-fix.sh

# Run type checking
./scripts/type-check.sh

# Run formatter
./scripts/code-format.sh

Acknowledgments

This project is inspired by mcp-server-mercurial.

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
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
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
VeyraX MCP

VeyraX MCP

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

Official
Featured
Local
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
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
E2B

E2B

Using MCP to run code via e2b.

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
Qdrant Server

Qdrant Server

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

Official
Featured