io.github.Murugarajr/ohm-mcp

Official

by Murugarajr · Python

AST-based Python refactoring MCP server with safe automated refactorings and rollback

Python code refactoring tool with AST analysis, safe automation, and rollback.

OHM-MCP

AI-Powered Python Refactoring & Code Quality Assistant

Works with GitHub Copilot, Cursor IDE, Cline, and any MCP-compatible AI assistant.

</div>

✨ Core Capabilities

🏗️ Architecture

God Object Detection
SOLID Violation Analysis
Design Pattern Suggestions
Dependency Injection Refactoring

🔧 Code Quality

AST Extract Method (100% accurate)
Dead Code Elimination
Import Refactoring
Symbol Renaming (project-wide)
Duplication Detection

</td> <td width="50%">

📊 Type Safety

Type Coverage Analysis
Type Stub Generation
Auto Test Generation

⚡ Performance

O(n²) Pattern Detection
Hotspot Analysis
Coverage-Driven Prioritization

🤖 Automation

Auto-Apply with Rollback
Operation History
Quality Dashboard

</td> </tr> </table>

🚀 Quick Start

Installation

📦 Recommended: NPM Method (Auto-installs dependencies)

No installation required! Use NPX to run the latest version automatically:

bash

npx -y ohm-mcp@latest

This is the easiest way to get started. Just add the configuration to your AI assistant (see IDE Configuration below).

MCP Registry: mcp-name: io.github.Murugarajr/ohm-mcp

🐍 Alternative: PyPI Method (For local development)

Use this method when you need to:

Develop locally with the Python package
Use a specific Python virtual environment
Install from source for customization

Install from PyPI:

bash

pip install ohm-mcp

Install from source (for development):

bash

pip install -e .

IDE Configuration

📦 Option 1: NPX (Recommended - Auto-installs dependencies)

<details> <summary>🔵 GitHub Copilot (VS Code) with NPX</summary>

After publishing to npm:

json

{
  "mcpServers": {
    "ohm-mcp": {
      "command": "npx",
      "args": ["ohm-mcp@latest"]
    }
  }
}

For local development:

json

{
  "mcpServers": {
    "ohm-mcp": {
      "command": "npx",
      "args": ["--package", "/path/to/ohm-mcp-npm", "ohm-mcp"]
    }
  }
}

Usage:

Open Copilot Chat
Type # and select ohm-mcp tools
Ask: "Analyze this file and suggest refactorings"

</details> <details> <summary>🟣 Cursor IDE with NPX</summary>

Global (After publishing to npm):

json

{
  "mcpServers": {
    "ohm-mcp": {
      "command": "npx",
      "args": ["-y", "ohm-mcp@latest"]
    }
  }
}

For local development:

json

{
  "mcpServers": {
    "ohm-mcp": {
      "command": "npx",
      "args": ["--package", "/path/to/ohm-mcp-npm", "ohm-mcp"]
    }
  }
}

Usage:

Open Cursor Chat (Cmd+L / Ctrl+L)
Tools are automatically available
Ask: "Use ohm-mcp to detect dead code"

</details> <details> <summary>🟢 Cline (VS Code Extension) with NPX</summary>

Global (After publishing to npm):

json

{
  "mcpServers": {
    "ohm-mcp": {
      "command": "npx",
      "args": ["-y", "ohm-mcp@latest"]
    }
  }
}

For local development:

json

{
  "mcpServers": {
    "ohm-mcp": {
      "command": "npx",
      "args": ["--package", "/path/to/ohm-mcp-npm", "ohm-mcp"]
    }
  }
}

Usage:

Open Cline panel
Tools are available in agent context
Ask: "Analyze type coverage and suggest improvements"

</details> <details> <summary>🟠 Antigravity with NPX</summary>

Global (After publishing to npm):

json

{
  "mcpServers": {
    "ohm-mcp": {
      "command": "npx",
      "args": ["-y", "ohm-mcp@latest"],
      "env": {
        "PYTHONUNBUFFERED": "1"
      }
    }
  }
}

Usage:

Tools are automatically available in Antigravity
Ask: "Use ohm-mcp to analyze architecture"

</details> <details> <summary>🔴 Roo Code with NPX</summary>

Global (After publishing to npm):

json

{
  "mcpServers": {
    "ohm-mcp": {
      "command": "npx",
      "args": ["-y", "ohm-mcp@latest"]
    }
  }
}

Usage:

Open Roo Code panel
Tools are available in agent context
Ask: "Use ohm-mcp to detect duplicates"

</details> <details> <summary>⚫ Codex VS Code with NPX</summary>

Configuration (config.toml):

toml

[mcp_servers.ohm-mcp]
args = ["-y", "ohm-mcp@latest"]
command = "npx"

Usage:

Open Codex panel
Tools are automatically available
Ask: "Use ohm-mcp to suggest design patterns"

</details> <details> <summary>🟡 Kilo Code with NPX (Local/Global)</summary>

Local (For local development with virtual environment):

json

{
  "mcpServers": {
    "ohm-mcp": {
      "command": "/Users/username/project/venv/bin/python",
      "args": ["-m", "ohm_mcp.server"],
      "disabled": false,
      "alwaysAllow": []
    }
  }
}

Global (After publishing to npm):

json

{
  "mcpServers": {
    "ohm-mcp": {
      "command": "npx",
      "args": ["-y", "ohm-mcp@latest"]
    }
  }
}

Usage:

Open Kilo Code panel
Tools are available in agent context
Ask: "Use ohm-mcp for refactoring"

</details>

🐍 Option 2: Direct Python (Manual setup)

<details> <summary>🔵 GitHub Copilot (VS Code) with Python</summary>

First install: pip install ohm-mcp

Then add to .vscode/mcp.json:

json

{
  "servers": {
    "ohm-mcp": {
      "command": "python",
      "args": ["-m", "ohm_mcp.server"]
    }
  },
  "inputs": []
}

Usage:

Open Copilot Chat
Type # and select ohm-mcp tools
Ask: "Analyze this file and suggest refactorings"

</details> <details> <summary>🟣 Cursor IDE with Python</summary>

First install: pip install ohm-mcp

Then add to Cursor's MCP settings file (.cursorrules or MCP config):

json

{
  "mcpServers": {
    "ohm-mcp": {
      "command": "python",
      "args": ["-m", "ohm_mcp.server"]
    }
  },
  "inputs": []
}

Example with virtual environment:

json

{
  "mcpServers": {
    "ohm-mcp": {
      "command": "/Users/username/projects/venv/bin/python",
      "args": ["-m", "ohm_mcp.server"]
    }
  }
}

Usage:

Open Cursor Chat (Cmd+L / Ctrl+L)
Tools are automatically available
Ask: "Use ohm-mcp to detect dead code"

</details> <details> <summary>🟢 Cline (VS Code Extension) with Python</summary>

First install: pip install ohm-mcp

Then add to Cline's MCP settings:

json

{
  "mcpServers": {
    "ohm-mcp": {
      "command": "python",
      "args": ["-m", "ohm_mcp.server"]
    }
  }
}

Example with virtual environment:

json

{
  "mcpServers": {
    "ohm-mcp": {
      "command": "/Users/username/projects/venv/bin/python",
      "args": ["-m", "ohm_mcp.server"]
    }
  }
}

Note: Cline requires absolute paths for both command and cwd.

Usage:

Open Cline panel
Tools are available in agent context
Ask: "Analyze type coverage and suggest improvements"

</details> <details> <summary>🔧 Other MCP-Compatible Clients</summary>

Any MCP-compatible client can use this server. General configuration:

json

{
  "mcpServers": {
    "ohm-mcp": {
      "command": "<python-interpreter-path>",
      "args": ["<path-to-mcp_server.py>"],
      "cwd": "<project-directory>"
    }
  }
}

Finding your Python path:

bash

# Unix/Mac
which python
# or
which python3

# Windows
where python

</details>

💡 Quick Usage Examples

Once configured, simply reference tools in your AI assistant chat using the format: use #ohm-mcp.tool_name on the current file or @file_name.py

🗑️ Find Dead Code

code

Use #ohm-mcp.detect_dead_code on @utils.py

This will detect:

✅ Unused imports (import statements never referenced)
✅ Unused variables (assigned but never read)
✅ Unreachable code (after return/raise/break/continue)
✅ Unused functions (defined but never called)
✅ Shadowed variables (inner scope hides outer scope variable)

📋 Code Duplication Detection

code

Use #ohm-mcp.detect_code_duplicates to find duplicates in /path/to/project

Finds:

✅ Exact duplicates (100% identical code blocks)
✅ Near duplicates (90%+ similarity)
✅ Duplicate functions (same structure, different names)
✅ Provides refactoring suggestions to eliminate duplication

🏗️ Architecture Analysis

code

Analyze architecture of @my_module.py using #ohm-mcp.analyze_architecture

Detects:

✅ God Objects (classes doing too much)
✅ SOLID principle violations
✅ Circular dependencies
✅ High coupling issues

✂️ Extract Method Refactoring

code

Use #ohm-mcp.extract_method_ast to extract lines 45-60 from @handler.py into a new function called "process_request"

Automatically:

✅ Detects required parameters
✅ Identifies return values
✅ Generates refactored code
✅ Creates unified diff patch

🔄 Safe Symbol Renaming

code

Use #ohm-mcp.rename_symbol to rename "old_function_name" to "new_function_name" in /path/to/project

Features:

✅ AST-based (100% accurate)
✅ Detects naming conflicts
✅ Shows all occurrences before applying
✅ Updates docstrings and comments

📊 Type Coverage Analysis

code

Analyze type hints in @module.py using #ohm-mcp.analyze_type_hints

Provides:

✅ Coverage percentage and grade
✅ Functions missing type hints
✅ Suggested type annotations
✅ Migration plan with priorities

⚡ Performance Optimization

code

Use #ohm-mcp.analyze_performance on @slow_module.py

Detects:

✅ Nested loops (O(n²) complexity)
✅ Quadratic list operations
✅ Repeated function calls (missing caching)
✅ Mutable default arguments
✅ Inefficient string concatenation

🧪 Auto-Generate Tests

code

Generate tests for @calculator.py using #ohm-mcp.generate_characterization_tests

Creates:

✅ Happy path test cases
✅ Edge cases (None, zero, negative, empty)
✅ Ready-to-run pytest code
✅ Preserves current behavior before refactoring

🎨 Design Pattern Suggestions

code

Suggest design patterns for @legacy_code.py using #ohm-mcp.suggest_design_patterns

Recommends:

✅ Strategy pattern for long if/elif chains
✅ Factory pattern for repetitive object creation
✅ Observer pattern for callback hell
✅ Decorator pattern for cross-cutting concerns

🔧 Import Refactoring

code

Use #ohm-mcp.refactor_imports to update all files in /path/to/project from "old.module" to "new.module"

Handles:

✅ Direct imports (import old.module)
✅ From imports (from old.module import X)
✅ Submodule imports
✅ Import aliases

🎯 Key Tools (30 Total)

<details> <summary>📋 General Analysis & Planning (4 tools)</summary>

Tool	Purpose	Output
`analyze_codebase`	Comprehensive code analysis	Issues + refactoring plan
`propose_function_refactor`	Function-level refactor planning	Detailed refactor proposal
`explain_refactoring`	Explain refactoring patterns	Educational guidance
`create_refactor_patch`	Generate unified diff patches	Patch file

</details> <details> <summary>🏗️ Architecture & Design (4 tools)</summary>

Tool	Purpose	Output
`analyze_architecture`	Detect God Objects, SOLID violations	Detailed issue report
`suggest_design_patterns`	Recommend patterns (Strategy, Factory, Observer)	Pattern suggestions + examples
`analyze_tight_coupling`	Find coupling issues	DI recommendations
`suggest_di_refactor`	Generate DI code	Before/after refactor

</details> <details> <summary>🔧 Code Quality & Refactoring (10 tools)</summary>

Tool	Purpose	Key Feature
`extract_method_ast`	Extract code into function	100% AST-based accuracy
`suggest_extractable_methods`	Find extractable blocks	Cohesion scoring
`detect_dead_code`	Find unused code	5 types of dead code
`refactor_imports`	Update imports project-wide	Safe module renaming
`refactor_single_file_imports`	Refactor imports in one file	Single file focus
`analyze_wildcard_imports`	Find wildcard imports	Explicit replacements
`rename_symbol`	Rename across codebase	Conflict detection
`detect_code_duplicates`	Find DRY violations	Exact + near duplicates
`extract_function_code`	Extract single function code	Code extraction utility
`apply_function_refactor`	Apply function-level refactor	Direct code modification

Example - Extract Method:

python

# Input: Lines 45-60
result = extract_method_ast(code, 45, 60, "calculate_total")

# Output: Refactored code + patch + auto-detected params/returns

</details> <details> <summary>📊 Type Safety & Testing (5 tools)</summary>

Tool	Purpose	Benefit
`analyze_type_hints`	Check type coverage	Migration plan
`generate_type_stub`	Create .pyi files	Gradual typing
`generate_characterization_tests`	Auto-generate tests	Safe refactoring
`generate_test_for_function`	Single function tests	Targeted testing
`suggest_tests`	Suggest test strategies	Test planning

</details> <details> <summary>⚡ Performance & Prioritization (2 tools)</summary>

Tool	Purpose	Detects
`analyze_performance`	Find bottlenecks	Nested loops, mutable defaults, O(n²)
`prioritize_by_coverage`	Risk-based prioritization	High-risk uncovered code

</details> <details> <summary>🤖 Automated Execution & History (4 tools)</summary>

mermaid

graph LR
    A[apply_refactoring] --> B{Dry Run?}
    B -->|Yes| C[Show Preview]
    B -->|No| D[Create Backup]
    D --> E[Apply Changes]
    E --> F{Run Tests}
    F -->|Pass| G[Success]
    F -->|Fail| H[Auto Rollback]
    H --> I[rollback_refactoring]

Tool	Purpose
`apply_refactoring`	Auto-apply refactoring with safety checks
`rollback_refactoring`	Rollback previous refactoring
`show_refactoring_history`	View refactoring audit trail
`cleanup_old_backups`	Clean up old backup files

Features:

✅ Automatic backup before changes
✅ Test execution validation
✅ Auto-rollback on failure
✅ Full audit trail with history
✅ Automatic backup cleanup

</details> <details> <summary>📈 Metrics & Reporting (1 tool)</summary>

generate_quality_report - Comprehensive dashboard in HTML/Markdown/JSON

Output Preview:

code

📊 Health Score: 85/100 (Good)
📁 Files: 47 | Lines: 12,450 | Tech Debt: 23 pts

📊 Type Coverage: 67%
🗑️ Dead Code: 8 imports, 12 variables, 3 functions
⚡ Performance: 4 nested loops, 2 mutable defaults
📋 Duplication: 3 exact, 5 near-duplicates

Visual Dashboard:

🎨 Circular health gauge
📊 Color-coded metrics (🟢🟡🔴)
📈 Trend tracking ready
🔗 CI/CD integration (JSON export)

</details>

💡 Common Workflows

1️⃣ Safe Refactoring

bash

generate_characterization_tests → pytest → extract_method_ast → pytest

2️⃣ Eliminate Duplication

bash

detect_code_duplicates → review suggestions → extract_method_ast

3️⃣ Type Migration

bash

analyze_type_hints → follow migration plan → generate_type_stub

4️⃣ Performance Optimization

bash

analyze_performance → prioritize_by_coverage → apply fixes

5️⃣ Module Refactoring

bash

refactor_imports(old="myapp.old", new="myapp.new") → review patches

6️⃣ Symbol Renaming

bash

rename_symbol(old="calc", new="calculate", preview_only=True) → apply

7️⃣ Quality Tracking

bash

generate_quality_report(format="html") → open dashboard → track trends

🎨 Visual Examples

Quality Dashboard Preview

code

┌─────────────────────────────────────────────┐
│  🔍 Code Quality Dashboard                  │
├─────────────────────────────────────────────┤
│                                             │
│       ╭─────────╮     📁 Overview           │
│       │   85    │     Files: 47             │
│       │  /100   │     Lines: 12,450         │
│       ╰─────────╯     Tech Debt: 23         │
│     Health Score                            │
│                                             │
├─────────────────────────────────────────────┤
│  📊 Type Coverage       ⚡ Performance       │
│  ████████░░ 67%         🔴 4 nested loops   │
│  120/180 typed          🟡 2 mutable args   │
├─────────────────────────────────────────────┤
│  🗑️ Dead Code           📋 Duplication      │
│  8 imports              3 exact             │
│  12 variables           5 near              │
│  3 functions                                │
└─────────────────────────────────────────────┘

Symbol Rename Preview

diff

# Before
- def calc(x, y):
-     return x + y
- result = calc(5, 3)

# After
+ def calculate_sum(x, y):
+     return x + y
+ result = calculate_sum(5, 3)

✅ 1 function renamed
✅ 3 call sites updated
✅ 0 conflicts detected

🧠 Design Principles

Principle	Implementation
🧪 Test-First	Auto-generate characterization tests before refactoring
↩️ Reversible	Every change = backup + rollback capability
🎯 AST-Driven	100% accurate (no regex)
📊 Risk-Aware	Coverage + complexity = prioritization
🏛️ SOLID	Detect violations + concrete fixes
🚫 No Blindness	Analyze → Plan → Validate

🔌 IDE Compatibility

<table> <tr> <td align="center" width="100%"> 🤖 Any MCP Client ✅ Standard Protocol ✅ Easy Setup ✅ Works with all MCP-compatible AI assistants </td> </tr> </table>

📊 Comparison

Feature	OHM MCP	Traditional Tools
Accuracy	100% AST	~70% Regex
Safety	Auto backup/rollback	Manual
Testing	Auto-generates	Manual
Automation	Full	Suggestions only
Dashboard	HTML/JSON/MD	Text logs
IDE Support	Copilot/Cursor/Cline	Limited

🎯 Use Cases

<table> <tr> <td>👨‍💻 Developers • Refactor legacy code safely • Find dead code • Optimize performance</td> <td>👥 Teams • Track tech debt • Enforce standards • Design patterns</td> <td>🚀 CI/CD • Quality gates • Trend tracking • Block bad PRs</td> </tr> </table>

📈 Metrics

code

✅ 30 MCP Tools
✅ 40+ Static Checks
✅ 100% AST Accuracy
✅ Zero Regex Patterns
✅ Automated Execution with Rollback
✅ Beautiful Dashboards (HTML/JSON/MD)
✅ Universal MCP Compatibility
✅ Safe Refactoring with Auto-Backup

🛠️ Troubleshooting

<details> <summary>MCP Connection Issues</summary>

Verify Python path:

bash

which python  # Unix/Mac
where python  # Windows

Test MCP server directly:
bash
```
python -m ohm_mcp.server
```
Check logs:
- VS Code: Check Output panel
- Cursor: Check Cursor logs
- Cline: Check Cline settings panel
Common issues:
- ❌ Relative paths in command → Use absolute paths
- ❌ Missing virtual environment → Activate venv first
- ❌ Wrong cwd for Cline → Must be absolute path

</details>

🤝 Contributing

Run before submitting:

bash

./static_analyser.sh  # Runs ruff, mypy, pylint, flake8
pytest                 # All tests must pass

🙏 Credits

Built with Model Context Protocol • Python AST • Compatible with GitHub Copilot, Cursor IDE, Cline

Made with ❤️ for better code quality

⭐ Star this repo if it helps you write cleaner code!

Documentation

</div>

Install

claude_desktop_config.json

{
  "mcpServers": {
    "ohm-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "ohm-mcp@1.0.1"
      ]
    }
  }
}