File Tools MCP Server
An MCP server that provides file and directory tools for LLMs, including project structure analysis, file reading, and project context resources, with support for .gitignore patterns.
README
File Tools MCP Server
An MCP (Model Context Protocol) server that provides file and directory tools for LLMs, with built-in support for respecting .gitignore patterns.
Features
Tools
- project_structure: Generates a tree view of directory structures while respecting
.gitignorepatterns - read_multiple_files: Reads multiple files in a single request with clear content separation
Resources
- project://info: Comprehensive project information (type, dependencies, scripts, git status)
- git://status: Detailed git repository status with file changes and recent commits
- dependencies://list: Lists all project dependencies by package manager
- config://list: Finds and lists all configuration files in the project
- files://recent: Shows the 20 most recently modified files
- context://full: Complete project context with directory structure and all text files under 50KB
Additional Features
- Handles circular symlinks and permission errors gracefully
- Supports depth limiting and hidden file visibility options
- Clear error messages with helpful context for path issues
Installation
For Claude Desktop
-
Install the server to Claude Desktop:
uv run mcp install main.py --name "File Tools" -
Restart Claude Desktop
For Development
- Clone the repository
- Install dependencies:
uv sync
Usage
Development Mode (with Inspector)
uv run mcp dev main.py
This opens the MCP Inspector for debugging and testing.
Standalone Mode
uv run python main.py
Runs the server in stdio mode, waiting for MCP client connections.
Testing
uv run python test_client.py
Runs a test client to verify server functionality.
Available Tools
project_structure
Generates a tree view of a directory structure while respecting .gitignore patterns.
Parameters:
path(str, optional): Directory path to analyze (default: current working directory)max_depth(int, optional): Maximum depth to traverse (default: None for unlimited)show_hidden(bool, optional): Include hidden files and directories (default: False)
Example:
# Get structure of current directory
result = await session.call_tool("project_structure", {})
# Get structure with depth limit
result = await session.call_tool("project_structure", {
"path": "/path/to/project",
"max_depth": 3
})
# Include hidden files
result = await session.call_tool("project_structure", {
"show_hidden": True
})
read_multiple_files
Reads multiple files and returns their contents in a single response.
Parameters:
paths(list[str], required): List of file paths to read
Features:
- Clear separation between files with headers showing path and size
- Graceful error handling for missing or inaccessible files
- Efficient for reading multiple files in one request
Example:
# Read multiple files
result = await session.call_tool("read_multiple_files", {
"paths": [
"src/main.py",
"config.json",
"README.md"
]
})
Available Resources
project://info
Provides comprehensive project information in one request:
- Project type detection (Node.js, Python, Rust, etc.)
- Package metadata (name, version, description)
- Available scripts/commands
- Dependency counts
- Git status summary
- Recent commits
git://status
Detailed git repository status:
- Current branch and tracking information
- Ahead/behind status with remote
- Staged, modified, and untracked files
- Last 10 commits
dependencies://list
Lists all project dependencies:
- Supports npm, yarn, pnpm, pip, poetry, uv
- Shows production and development dependencies
- Handles multiple package manager formats
config://list
Finds all configuration files:
- Environment files (.env)
- Language configs (tsconfig.json, pyproject.toml)
- Linters and formatters
- CI/CD configurations
- Editor settings
files://recent
Shows recently modified files:
- 20 most recently modified files
- File modification timestamps
- File sizes
- Excludes common ignore patterns
context://full
Complete project context in one request:
- Full directory structure
- Contents of all text files under 50KB
- Respects .gitignore patterns
- Skips binary files automatically
- Provides summary of files included/skipped
- Perfect for giving LLMs complete project understanding
Configuration
Root Paths
When using the MCP Inspector or other MCP clients, you may need to configure root paths to allow the server to access your project directories. This is a security feature that prevents unauthorized file access.
In the MCP Inspector:
- Click on the server configuration
- Add root paths for directories you want to access
- Example:
/home/user/projects/my-project
Without proper root paths configured, you may see "File not found" errors even for files that exist.
Security Considerations
When deploying this server:
- Bind to localhost (127.0.0.1) only for local deployments
- Use proper authentication if exposing to network
- Consider path traversal risks - the server allows reading any accessible directory
- Configure root paths to limit file access to specific directories
License
MIT
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。