rope-mcp-server
Provides Python refactoring capabilities via the Rope library, enabling AI agents to perform safe, project-wide code transformations such as renaming symbols, moving modules, and extracting methods.
README
rope-mcp-server
A Model Context Protocol (MCP) server that provides Python refactoring capabilities powered by the Rope library. Enables AI agents like Claude to perform safe, project-wide refactoring operations.
Features
- Move Symbol - Move classes and functions between files with automatic import updates
- Move Module - Move entire modules or packages to different folders
- Move and Rename Module - Move module to folder and rename (e.g.,
foo_extra.py→foo/extra.py) - Convert Module to Init - Transform
foo.pyintofoo/__init__.py(no import changes!) - Convert Module to Package - Transform
foo.pyintofoo/foo.pywith import updates - Rename Symbol - Rename variables, functions, classes across entire projects
- Extract Method - Extract code blocks into new methods
- Inline Variable - Inline variables at all usage sites
- List Symbols - Discover top-level symbols in Python files
All operations are project-aware and automatically update imports and references throughout your codebase.
Installation
Quick Start with Claude Code (recommended)
# Global installation (all projects)
claude mcp add rope-refactor uvx rope-mcp-server -s user
# Or for current project only
claude mcp add rope-refactor uvx rope-mcp-server -s local
From PyPI
pip install rope-mcp-server
Usage with Claude Code
claude mcp add rope-refactor uvx rope-mcp-server -s user
claude mcp list # verify
That's it.
Available Tools
list_symbols
List all top-level symbols (classes, functions, variables) in a Python file.
project_path: /path/to/your/project
file_path: src/models.py
move_symbol
Move a class or function to another file. Automatically updates all imports.
project_path: /path/to/your/project
source_file: src/models.py
symbol_name: UserModel
dest_file: src/users/models.py
move_module
Move a module or package to another folder. Automatically updates all imports.
project_path: /path/to/your/project
module_path: src/utils.py
dest_folder: src/lib
convert_module_to_init (recommended)
Convert a module file into a package by moving it to __init__.py. No import changes needed!
This is the recommended way to convert a module to a package. After conversion, you can use move_module to move related files (like foo_mixins.py) into the new package.
project_path: /path/to/your/project
module_path: app/views/customer_order.py
Before:
app/views/customer_order.py
app/views/customer_order_mixins.py
After running convert_module_to_init:
app/views/customer_order/__init__.py (was customer_order.py)
app/views/customer_order_mixins.py (move this next with move_module)
Import from app.views.customer_order import X stays unchanged!
move_and_rename_module
Move a module to a folder and optionally rename it. Perfect for moving related files (like _mixins.py, _extra.py) into a package created with convert_module_to_init.
Auto-detection: If the module name starts with the destination folder name + underscore, it strips that prefix automatically.
Rope bug workaround: This tool includes a workaround for a Rope bug that crashes when a file has imports from BOTH the destination package AND the module being moved. Such files are temporarily hidden during the move, then their imports are fixed manually.
project_path: /path/to/your/project
module_path: app/views/customer_order_mixins.py
dest_folder: app/views/customer_order
new_name: mixins # Optional - auto-detected from prefix
Before:
app/views/customer_order/__init__.py
app/views/customer_order_mixins.py
Import: from app.views.customer_order_mixins import MyMixin
After:
app/views/customer_order/__init__.py
app/views/customer_order/mixins.py
Import: from app.views.customer_order.mixins import MyMixin
convert_module_to_package
Convert a module file into a package with the same name. Transforms foo.py into foo/foo.py while updating all imports project-wide.
Use this when you want the original module content in a submodule, not in __init__.py.
project_path: /path/to/your/project
module_path: app/views/service_contractor.py
Before:
app/views/service_contractor.py
Import: from app.views.service_contractor import MyClass
After:
app/views/service_contractor/
├── __init__.py
└── service_contractor.py
Import: from app.views.service_contractor.service_contractor import MyClass
rename_symbol
Rename a symbol across the entire project.
project_path: /path/to/your/project
file_path: src/utils.py
symbol_name: old_function_name
new_name: new_function_name
extract_method
Extract a code region into a new method.
project_path: /path/to/your/project
file_path: src/service.py
start_line: 15
start_col: 4
end_line: 20
end_col: 30
new_name: extracted_helper
inline_variable
Inline a variable at all usage sites.
project_path: /path/to/your/project
file_path: src/handler.py
variable_name: temp_result
line: 42
close_rope_project
Close a Rope project to free memory. Call when done with refactoring.
project_path: /path/to/your/project
Development
# Install dependencies
uv sync
# Run tests
uv run pytest tests/ -v
# Run server locally
uv run python -m rope_mcp_server.server
How it works
This server wraps the Rope refactoring library and exposes its capabilities via the Model Context Protocol. When Claude (or any MCP-compatible agent) needs to refactor Python code, it can use these tools to perform safe, AST-aware transformations that preserve code correctness.
Key implementation details:
- Project caching - Rope projects are cached to avoid re-parsing on every operation
- AST-based offset calculation - Symbol locations are computed via Python's AST module for accuracy
- Automatic import handling - Rope handles all import updates when moving/renaming symbols
Limitations
- Only supports Python code (Rope limitation)
- Moving methods between classes requires the target class to exist
- Large projects may have slower initial indexing
- Rope bug workaround: Files with imports from both the destination package and the module being moved are handled specially (hidden during move, then imports fixed manually) - see
move_and_rename_module
License
MIT
Contributing
Contributions welcome! Please open an issue first to discuss what you would like to change.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。