jetbrains-proxy-mcp-server
A proxy server that forwards MCP requests to a JetBrains IDE server, with support for path conversion between WSL, Git Bash, and Windows.
README
JetBrains Proxy MCP Server
Project Overview:
- A Python-based proxy server for the JetBrains MCP (Model Context Protocol) server that acts as an intermediary between an MCP client and a JetBrains MCP server.
- The core logic resides in
src/jetbrains_proxy_mcp_server/server.py, which handles the proxy server setup and transport layer. Thesrc/jetbrains_proxy_mcp_server/service/JetbrainsMCPServerProxy.pyclass manages the connection to the downstream JetBrains MCP server and proxies the tool calls. It also includes logic to handle path conversions between different path formats (WSL, Git Bash, Windows).
Configuration is managed through the src/jetbrains_proxy_mcp_server/properties/MCPServerProperties.py class,
which loads settings from a config.yaml file.
Features:
- Proxy Functionality: Forwards requests from MCP clients to the JetBrains MCP server
- Dual Transport Support: Works with both Server-Sent Events (SSE) and standard I/O (stdio)
- Path Conversion: Handles file path conversions between different path formats (WSL, Git Bash, Windows)
- Resilience: Built-in retry, timeout, and restart mechanisms for robust operation
- Tool Filtering: Supports a curated list of JetBrains tools for controlled access
Installation
See pytools.sh
Usage: pytools.sh <command> [args...]
Commands:
test [pytest-args...] Always run pytest in project .venv (auto-create). If pytest missing, auto-install.
purge Remove temp/build files (.venv, build, dist, caches, *.egg-info, _version.py)
reinstall-system [pip-args...] Reinstall into SYSTEM Python. Pass extra args to 'pip install'. No purge.
Examples:
pytools.sh reinstall-system --break-system-packages
pytools.sh reinstall-system --no-build-isolation ".[dev]"
reinstall-venv [pip-args...] Reinstall into project .venv (auto-create). Pass extra args to 'pip install'. No purge.
Examples:
pytools.sh reinstall-venv
pytools.sh reinstall-venv --no-build-isolation ".[dev]"
upload <repository> [extra-pip-args...] Build (sdist+wheel) and upload via twine to the named repository (e.g. pypi, testpypi).
Pass extra args to 'pip install' after the default list.
Examples:
pytools.sh upload pypi
pytools.sh upload testpypi 'build==1.2.2' 'twine==5.0.0'
Env vars:
PYTHON_BIN Python command to use (default: python3)
VENV_DIR Virtualenv path (default: $script_dir/.venv)
# Install the package using the provided script:
./pytools.sh reinstall-venv
./pytools.sh reinstall-system --break-system-packages
# Or install the package using pip:
pip install .
Configuration
See also:
Create a config.yaml file to configure the proxy server. The server will look for this file in the following order:
- Path specified by the
--configcommand line argument config.yamlin the current directory~/.config/jetbrains-proxy-mcp-server/config.yaml
Example Configuration
# Server configuration
server-name: "JetBrains Proxy MCP Server"
transport: "sse" # or "stdio"
timeout: 60.0
sse-port: 41110
sse-debug-enabled: true
# JetBrains MCP server configuration
jetbrains-mcp-server:
name: "jetbrains-mcp-server"
url: "http://127.0.0.1:64342/sse"
timeout: 35.0
sse-read-timeout: 300.0
start-timeout: 120.0
stop-timeout: 30.0
max-attempts: 5
initial-backoff: 1.0
max-backoff: 60.0
backoff-multiplier: 3.0
debug-enabled: true
client-path-type: "wsl" # or ""windows_git_bash" or "windows"
server-path-type: "windows" # or "wsl" or "windows_git_bash"
Proxy Server Properties
| Property | Default | Description |
|---|---|---|
server-name |
JetBrains Proxy MCP Server | Name of the proxy server |
transport |
sse | Transport type: "sse" or "stdio" |
sse-port |
41110 | Port for SSE server |
sse-debug-enabled |
true | Enable SSE debug mode |
timeout |
60.0 | Timeout for tool calls in seconds |
JetBrains MCP Server Properties
| Property | Default | Description |
|---|---|---|
name |
jetbrains-mcp-server | Name of the JetBrains MCP server |
url |
http://127.0.0.1:64342/sse | URL of the JetBrains MCP server |
headers |
{} | Additional headers for requests |
timeout |
35.0 | Timeout for requests to JetBrains server |
sse-read-timeout |
300.0 | SSE read timeout |
start-timeout |
120.0 | Timeout for server startup |
stop-timeout |
30.0 | Timeout for server shutdown |
max-attempts |
5 | Maximum retry attempts |
initial-backoff |
1.0 | Initial backoff time in seconds |
max-backoff |
60.0 | Maximum backoff time in seconds |
backoff-multiplier |
3.0 | Backoff multiplier for retries |
client-path-type |
wsl | Path type for the client |
server-path-type |
windows | Path type for the JetBrains server |
debug-enabled |
true | Enable debug logging |
Environment Variables
See also:
Configuration can also be set using environment variables:
JETBRAINS_PROXY_MCP_SERVER_CONFIG- Path to config fileJETBRAINS_PROXY_MCP_SERVER_NAME- Server nameJETBRAINS_PROXY_MCP_SERVER_TRANSPORT- Transport typeJETBRAINS_PROXY_MCP_SERVER_TIMEOUT- Tool timeout
For stdio transport, you must set:
export SIMP_LOGGER_LOG_CONSOLE_ENABLED=False
MCP Configuration
JSON
sse, wsl:
{
"mcpServers": {
"jetbrains": {
"url": "http://127.0.0.1:41110/sse"
}
}
}
sse, windows:
{
"mcpServers": {
"jetbrains": {
"url": "http://127.0.0.1:41111/sse"
}
}
}
stdio, windows:
pip install jetbrains-proxy-mcp-server --upgrade --force-reinstall --extra-index-url http://127.0.0.1:8081/repository/pypi-group/simple --trusted-host 127.0.0.1
{
"mcpServers": {
"jetbrains": {
"command": "jetbrains-proxy-mcp-server",
"args": [
"--config",
"c:/Users/<USER>/.config/jetbrains-proxy-mcp-server/config.yaml"
],
"env": {
"SIMP_LOGGER_LOG_CONSOLE_ENABLED": "False",
"SIMP_LOGGER_LOG_LEVEL": "DEBUG"
}
}
}
}
Running the Server
Start the server with:
jetbrains-proxy-mcp-server
# Or:
jetbrains-proxy-mcp-server --config /path/to/your/config.yaml
If no config path is provided, the server will search for a config.yaml file in the default locations.
Supported Tools
The proxy supports a curated list of JetBrains tools:
MCP Server v252.xxx
get_all_open_file_paths- Get paths of all open filesget_file_problems- Analyze file for errors and warningsget_file_text_by_path- Retrieve text content of a fileget_project_dependencies- Get project dependenciesget_project_modules- Get project modulesget_project_problems- Get project-wide problemslist_directory_tree- List directory contents in tree formatreformat_file- Reformat a filerename_refactoring- Rename a symbol across the projectreplace_text_in_file- Replace text in a filesearch_in_files_by_regex- Search files using regexsearch_in_files_by_text- Search files for text
Path Conversion
The proxy handles path conversions between different operating systems:
- WSL:
/mnt/c/path/to/file - Git Bash:
/c/path/to/file - Windows:
C:\path\to\file
Configure client-path-type and server-path-type in your config to match your environments.
Development
Development Conventions
- The project uses
setuptoolsfor packaging. - Dependencies are managed in
pyproject.toml. - The code follows standard Python conventions.
- Logging is implemented using the
loggingmodule. - The project uses
anyiofor asynchronous operations. - The server uses
uvicornwhen running with the SSE transport.
Installing dependencies
To install the necessary dependencies, run:
uv sync --no-install-project --extra test
Testing
Run tests with pytest:
./pytools test tests/test_xxx.py
# Or use the command directly:
pytest tests/test_xxx.py
# Or test all:
pytest
Dependencies
- Python >= 3.10
- mcp >= 1.4.0
- pydantic >= 2.0
- typing >= 3.7.4.3
- pyyaml >= 6.0.2
- anyio >= 4.3.0
License
This project is licensed under the MIT License. See the LICENSE file for details.
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。