发现优秀的 MCP 服务器

Excel MCP Server

一个模型上下文协议服务器，使 AI 助手能够读取和写入 Microsoft Excel 文件，支持诸如 xlsx、xlsm、xltx 和 xltm 等格式。

Crawlab MCP Server

Nexus MCP Bridge for VSCode

一个轻量级的桥接扩展，使 Claude Desktop 能够通过模型上下文协议连接到 VSCode 工作区，允许 Claude 读取和写入文件、创建目录以及列出工作区中的内容。

Windows CLI MCP Server

一个模型上下文协议（Model Context Protocol，MCP）服务器，提供对 Windows 系统的安全命令行访问。它允许像 Claude Desktop 这样的 MCP 客户端安全地在 PowerShell、CMD 和 Git Bash shell 中执行命令，并具有可配置的安全控制。

SourceSage MCP

一个基于 TypeScript 的服务器，可以将项目目录结构可视化为 Markdown 格式，自动记录文件内容并进行语法高亮显示，同时支持自定义排除模式。

Markdownify MCP Server - UTF-8 Enhanced

一个文档转换服务器，可以将各种文件格式（PDF、文档、图像、音频、网页内容）转换为 Markdown，并改进了对多语言和 UTF-8 的支持。

MCP-Delete

一个模型上下文协议（MCP）服务器，提供文件删除功能。该服务器允许 AI 助手在需要时安全地删除文件，并支持相对路径和绝对路径。

Excel Reader Server

一个模型上下文协议 (MCP) 服务器，提供读取 Excel (xlsx) 文件的工具，能够从整个工作簿或特定工作表中提取数据，并将结果以结构化的 JSON 格式返回。

MCP Python Toolbox

一个模型上下文协议服务器，使像 Claude 这样的人工智能助手能够通过文件操作、代码分析、项目管理和安全的代码执行来执行 Python 开发任务。

WASM MCP Server

这个服务器支持在网页浏览器中运行模型上下文协议，其功能包括算术运算和基于会话的键值存储。

EverArt Forge MCP Server

一个为 Cline 设计的高级 MCP 服务器，它利用 EverArt 的 AI 模型来生成矢量和栅格图像，支持灵活的存储、多种格式以及强大的图像生成能力。

Google Drive MCP Server

与 Google Drive 集成，以实现文件列表、搜索和读取，以及 Google 表格的读取和写入。

File Operations MCP Server

一个模型上下文协议服务器，它支持增强的文件系统操作，包括具有流式传输功能的读取、写入、复制、移动文件，以及目录管理、文件监视和变更跟踪。

Obsidian MCP Server

通过模型上下文协议，实现LLM与Obsidian库之间的交互，支持安全的文件操作、内容管理和高级搜索功能。

Filesystem MCP Server

一个 MCP 服务器，允许 Claude AI 执行文件系统操作，包括在指定的允许路径内读取、写入、列出、移动文件和搜索目录。 (Alternative, slightly more formal and technical:) 一个 MCP 服务器，它赋予 Claude AI 在预先设定的允许路径下执行文件系统操作的能力，包括读取、写入、列出、移动文件以及搜索目录。

MCP Source Tree Server

Okay, I understand. Here's how you can generate a JSON file tree from a specified directory's `src` folder, respecting `.gitignore` rules, suitable for quick project structure review in Claude. I'll provide a Python script to accomplish this. **Explanation:** 1. **`gitignore_parser` Library:** We'll use the `gitignore_parser` library to correctly interpret `.gitignore` files. This is crucial for accurately reflecting what files should be excluded. If you don't have it, you'll need to install it: `pip install gitignore_parser` 2. **`os.walk()`:** This function recursively traverses the directory tree. 3. **`.gitignore` Handling:** The script reads and parses `.gitignore` files in each directory it encounters. 4. **JSON Output:** The script constructs a JSON representation of the file tree. **Python Script:** ```python import os import json import gitignore_parser def generate_file_tree_json(root_dir, output_file="file_tree.json"): """ Generates a JSON file tree from the 'src' folder of a specified directory, respecting '.gitignore' rules. Args: root_dir (str): The root directory of the project. output_file (str): The name of the output JSON file. Defaults to "file_tree.json". """ src_dir = os.path.join(root_dir, "src") if not os.path.exists(src_dir): print(f"Error: 'src' directory not found in {root_dir}") return def build_tree(directory, ignore_checker): """Recursively builds the file tree.""" tree = {} for item in os.listdir(directory): full_path = os.path.join(directory, item) relative_path = os.path.relpath(full_path, src_dir) # Path relative to src if ignore_checker(relative_path): continue # Skip ignored files/directories if os.path.isfile(full_path): tree[item] = None # Represent files as None elif os.path.isdir(full_path): tree[item] = build_tree(full_path, ignore_checker) return tree def create_ignore_checker(root_directory): """Creates a function to check if a file/directory is ignored based on .gitignore files.""" ignore_files = [] for root, _, files in os.walk(root_directory): if '.gitignore' in files: ignore_files.append(os.path.join(root, '.gitignore')) ignore_list = [] for ignore_file in ignore_files: ignore_list.append(gitignore_parser.parse(ignore_file)) def is_ignored(path): for ignore in ignore_list: if ignore(os.path.join(src_dir, path)): # Check against the full path return True return False return is_ignored ignore_checker = create_ignore_checker(src_dir) file_tree = build_tree(src_dir, ignore_checker) with open(output_file, "w") as f: json.dump(file_tree, f, indent=4) print(f"File tree JSON saved to {output_file}") # Example Usage: if __name__ == "__main__": # Replace with the actual root directory of your project project_root = "/path/to/your/project" # <---- CHANGE THIS! generate_file_tree_json(project_root) ``` **How to Use:** 1. **Install `gitignore_parser`:** `pip install gitignore_parser` 2. **Replace Placeholder:** In the `if __name__ == "__main__":` block, replace `"/path/to/your/project"` with the actual absolute path to the root directory of your project (the directory containing the `src` folder). 3. **Run the Script:** Execute the Python script. It will create a file named `file_tree.json` in the same directory as the script. 4. **Upload to Claude:** Upload the `file_tree.json` file to Claude. **Example `file_tree.json` Output (Illustrative):** ```json { "components": { "Button.js": null, "Input.js": null }, "utils": { "api.js": null, "helpers.js": null }, "App.js": null, "index.js": null } ``` **Key Improvements and Considerations:** * **`.gitignore` Parsing:** Uses `gitignore_parser` for accurate `.gitignore` handling. This is *essential* for real-world projects. * **Error Handling:** Includes a check to ensure the `src` directory exists. * **Relative Paths:** Uses `os.path.relpath` to store paths relative to the `src` directory in the JSON, making the output more concise and readable. * **Clearer Structure:** Represents files as `null` in the JSON tree, which is a common and easily understood convention. * **Example Usage:** Provides a clear example of how to use the script. * **Comments:** Includes comments to explain the code. * **`create_ignore_checker` Function:** This function encapsulates the logic for creating the ignore checker, making the code more modular and readable. It finds all `.gitignore` files within the `src` directory and its subdirectories. * **Full Path Checking:** The `is_ignored` function now checks the full path (relative to the `src` directory) against the `.gitignore` rules, ensuring accurate matching. * **Modularity:** The code is broken down into functions for better organization and reusability. **How to Use with Claude:** 1. **Upload the JSON:** Upload the generated `file_tree.json` file to Claude. 2. **Prompt Claude:** Craft a prompt that asks Claude to analyze the file structure. For example: * "Here is a JSON representation of the file structure of a project's `src` directory. Can you identify the main components and utilities?" * "Analyze this file structure and suggest potential areas for refactoring." * "Based on this file structure, what design patterns might be in use?" * "This JSON represents the file tree of a React project. Identify the likely component structure." The more specific your prompt, the better the results you'll get from Claude. You can also ask Claude to generate diagrams or visualizations based on the JSON data.

Everything Search MCP Server

提供与 Everything 搜索器的集成，通过模型上下文协议实现强大的文件搜索功能，并提供高级搜索选项，如正则表达式、区分大小写和排序。

Deskaid

一个 MCP 服务器，提供用于在本地文件系统上读取、写入和编辑文件的工具。