Skip to main content
Glama

MCP Worktree Voting Server

by Doctacon
GitHub
Python
  • Apple
OverviewInspectNewEndpointsSchemaRelated ServersReviewsScore
Need Help?View Source CodeReport Issue

MCP Git Worktree Workflows

An MCP (Model Context Protocol) server that implements automated parallel development using git worktrees.

🚀 Key Features

  • Voting Workflow: Intelligently ranks and selects the highest-quality implementation

  • Ad Hoc Workflow: Quick single-worktree creation for simple tasks

  • Orchestration Workflow: Break complex tasks into subtasks with coordinated execution

  • Clean Workflow: Automatic cleanup of unsuccessful variants

Installation

  1. Install dependencies:

uv add fastmcp # or pip install fastmcp

  1. Add the server to Claude Code:

Option A: Using Claude Code CLI (Recommended)

# For project-specific use claude mcp add worktree-workflows python /path/to/mcp-servers/worktree_workflows.py # For global use across all projects claude mcp add --scope user worktree-workflows python /path/to/mcp-servers/worktree_workflows.py

Option B: Manual Configuration

Add to your MCP configuration file:

{ "mcpServers": { "worktree-workflows": { "command": "python", "args": ["/path/to/mcp-servers/worktree_workflows.py"] } } }

  1. Restart Claude Code or use /mcp command to reconnect

🔄 Workflow Options

1. Voting Pattern (Multiple Implementations)

create_voting_worktrees( task="Rewrite the README.md with better examples", num_variants=5, target_repo="lombardi" # Optional: specify repository )

2. Ad Hoc Single Worktree

create_adhoc_worktree( task="Fix the login bug in auth.js", target_repo="my-app" # Optional )

3. Orchestrated Subtasks

create_orchestrated_worktrees( task="Implement user authentication system", subtasks=[ "Create database models for users and sessions", "Build authentication API endpoints", "Implement JWT token generation and validation", "Create login/logout UI components", "Add authentication middleware and route protection" ], target_repo="my-app" # Optional )

📊 Evaluation Metrics

The system automatically evaluates implementations using:

  • Code Changes (30 points): Has meaningful modifications

  • Test Success (50 points): Tests pass successfully

  • File Impact (up to 20 points): Number of files modified

  • Quality Heuristics: Additional scoring based on implementation patterns

🛠️ Available Tools

Core Voting Workflow

  • create_voting_worktrees: Creates worktrees and starts automated execution

  • list_sessions: Monitor all active sessions and their progress

  • evaluate_implementations: Get detailed ranking and evaluation of all variants

  • auto_select_best: Automatically choose and finalize the best implementation

  • finalize_best: Manually select a specific implementation

  • cleanup_session: Force cleanup of sessions

Additional Workflows

  • create_adhoc_worktree: Single worktree for quick tasks

  • create_orchestrated_worktrees: Multiple worktrees for coordinated subtasks

Utility Functions

  • get_worktree_info: Inspect specific worktree details

  • mark_implementation_complete: Manually mark implementations as done

💡 Example Use Cases

Voting Pattern

Perfect for:

  • Architecture Exploration: Try different design patterns simultaneously

  • Library Comparison: Implement with various frameworks/libraries

  • Algorithm Optimization: Test multiple approaches to performance problems

  • UI/UX Variants: Create different interface implementations

Ad Hoc Tasks

Great for:

  • Bug Fixes: Quick isolation and resolution

  • Small Features: Rapid implementation without overhead

  • Experiments: Try ideas without affecting main branch

Orchestrated Development

Ideal for:

  • Large Features: Break down and parallelize development

  • System Refactoring: Coordinate multiple related changes

  • API Development: Build endpoints, models, and tests in parallel

🎯 Quick Start Examples

Example 1: Voting Pattern

# 1. Start automated voting session create_voting_worktrees( task="Add Redis caching to the user service with error handling", num_variants=3 ) # 2. Check progress (Claude is working automatically) list_sessions() # Returns: "2/3 complete, 3/3 executed" # 3. View results and rankings evaluate_implementations(session_id="xyz789") # Shows ranked implementations with scores # 4. Auto-select winner and merge auto_select_best(session_id="xyz789", merge_to_main=true) # Merges best implementation, cleans up others

Example 2: Quick Fix

# Create single worktree for a bug fix create_adhoc_worktree( task="Fix null pointer exception in user profile component" ) # Claude starts immediately in new Terminal window

Example 3: Feature Development

# Break down complex feature into subtasks create_orchestrated_worktrees( task="Add real-time notifications", subtasks=[ "Set up WebSocket server infrastructure", "Create notification database schema and models", "Build notification API endpoints", "Implement frontend notification component", "Add notification preferences to user settings" ] ) # 5 Terminal windows open, each with Claude working on a subtask

⚡ Performance Notes

  • Concurrent Execution: All Claude instances run in parallel

  • Automatic Cleanup: Failed/low-quality implementations are removed

  • Resource Efficient: Only keeps the winning implementation

  • Fast Evaluation: Uses git diff stats and automated test detection

  • Smart Naming: Worktrees include task description for easy identification

🔧 Advanced Features

Worktree Naming

  • Branches and directories now include task descriptions

  • Format: {session-id}-{task-slug}-{variant}

  • Example: abc123-fix-login-bug-var1

Origin/Main Support

  • Ad hoc worktrees always branch from origin/main

  • Ensures clean starting point for isolated tasks

  • Automatically fetches latest changes

Terminal Integration

  • Automatically spawns Terminal windows (macOS)

  • Each worktree gets its own Claude session

  • No manual terminal management required

This server cannot be installed

-
security - not tested
F
license - not found
-
quality - not tested

How are these scores calculated?

local-only server

The server can only run on the client's local machine because it depends on local resources.

Enables parallel implementation of tasks using git worktrees, allowing you to create multiple variants of a solution, evaluate them side-by-side, and select the best one.

  1. 🚀 Key Features
    1. Installation
      1. Option A: Using Claude Code CLI (Recommended)
      2. Option B: Manual Configuration
    2. 🔄 Workflow Options
      1. 1. Voting Pattern (Multiple Implementations)
      2. 2. Ad Hoc Single Worktree
      3. 3. Orchestrated Subtasks
    3. 📊 Evaluation Metrics
      1. 🛠️ Available Tools
        1. Core Voting Workflow
        2. Additional Workflows
        3. Utility Functions
      2. 💡 Example Use Cases
        1. Voting Pattern
        2. Ad Hoc Tasks
        3. Orchestrated Development
      3. 🎯 Quick Start Examples
        1. Example 1: Voting Pattern
        2. Example 2: Quick Fix
        3. Example 3: Feature Development
      4. ⚡ Performance Notes
        1. 🔧 Advanced Features
          1. Worktree Naming
          2. Origin/Main Support
          3. Terminal Integration

        Related MCP Servers

        • mcp-github-project-manager

          kunwarVivek
          A
          security
          A
          license
          A
          quality
          Provides comprehensive tools for managing GitHub projects, milestones, tasks, and sprints. This server integrates deeply with GitHub Projects V2, offering features like automated kanban workflows, sprint planning, and custom field management.
          Last updated -
          46
          15
          68
          MIT License

        • GitHub MCP Server

          asbloom-py
          -
          security
          F
          license
          -
          quality
          Enables interaction with GitHub through the GitHub API, supporting file operations, repository management, advanced search, and issue tracking with comprehensive error handling and automatic branch creation.
          Last updated -
          1,400
          1

        • GitHub MCP Server

          tuanle96
          A
          security
          F
          license
          A
          quality
          Enables comprehensive GitHub operations through natural language including file management, repository administration, issue tracking, and advanced code searching.
          Last updated -
          47
          0
          1

        • mcp-git

          ver0-project
          A
          security
          A
          license
          A
          quality
          Provides comprehensive Git operations as tools for AI assistants and applications. This server enables AI systems to interact with Git repositories, allowing to initialize, fetch, commit, log, status, etc..
          Last updated -
          10
          0
          1
          MIT License

        View all related MCP servers

        New MCP Servers

        MCP directory API

        We provide all the information about MCP servers via our MCP API.

        curl -X GET 'https://glama.ai/api/mcp/v1/servers/Doctacon/mcp-servers'

        If you have feedback or need assistance with the MCP directory API, please join our Discord server