The ZIP MCP Server acts as a digital packing assistant for AI models, allowing them to manage compressed files just like a human would. This tool gives AI assistants the ability to create new ZIP archives, open existing ones, and see exactly what's inside a folder without having to extract everything first. It is perfect for users who want their AI to help organize cluttered directories or bundle a batch of documents for easy sharing. Moving beyond basic file management, this server provides a robust set of parameters for fine-tuned control over archives. It supports multi-file packaging and offers specific controls for compression levels ranging from 0 to 9. For users handling sensitive information, the tool includes advanced security features such as password protection and adjustable encryption strength, ensuring that the AI can handle secure data workflows while maintaining file integrity. For developers integrating with LLM systems like Claude or Cursor, this MCP tool bridges the gap between high-level reasoning and low-level file system operations. By using the metadata retrieval tool, an AI can analyze the structure, compression ratio, and file sizes of an archive before deciding how to process the contents. This makes it an essential utility for building automated workflows where an AI must manage local storage, handle large datasets, or package software assets efficiently through a standardized protocol.
Category: Developer Tools & Code Intelligence
Tags: archiving, compression, encryption, file-management, zip
bash npm install -g zip-mcpmcpServers configuration JSON: json { "mcpServers": { "zip-mcp": { "command": "zip-mcp", "args": [] } } }compress: Compresses local files or directories into a ZIP file. * Parameters: * input (string or string array): Path of the file or directory to be compressed. * output (string): Path of the output ZIP file. * options (optional): level (0-9), password, encryptionStrength (1-3), overwrite (boolean). * decompress: Decompresses local ZIP files to a specified directory. * Parameters: * input (string): Path of the ZIP file. * output (string): Path of the output directory. * options (optional): password, overwrite (boolean), createDirectories (boolean). * getZipInfo: Retrieves metadata from local ZIP files. * Parameters: * input (string): Path of the ZIP file. * options (optional): password. * echo: A test tool to verify if the service is running correctly. * Parameters: * message (string): The message to be returned./path/to/files into a ZIP file named output.zip with a compression level of 9 and set the password to 'secret'." Decompressing an archive: "Extract all files from archive.zip to the directory /path/to/extract, ensuring that any existing files are overwritten." Checking ZIP contents: "Show me the metadata and file list for the ZIP file located at /path/to/archive.zip." Testing the server: "Send an echo message 'Hello ZIP MCP' to verify the server is active."compress tool to identify relevant files and bundle them into a single ZIP file with a high compression level (9). It can even exclude unnecessary folders like node_modules or .git by only selecting the necessary input paths. Example: "Claude, please bundle all the .ts files in my src folder and the README.md into a file named v1-alpha-release.zip using the highest compression level."error-logsdirectory intosecure-logs.zip`, set the encryption strength to 3, and use a strong random password."
getZipInfo tool, the AI can "peek" inside the archive to list the total number of files, the compression ratio, and the specific file names and sizes without actually decompressing the file. Example: "I just downloaded dataset.zip. Before I extract it, tell me what files are inside, how big they are, and if the compression ratio suggests it contains mostly text or binary data."compress tool to archive…Part of MCP Servers
ZIP MCP Server allows AI assistants to compress files or directories, extract existing ZIP archives, and retrieve detailed metadata without extracting files. It supports multi-file selection, compression levels from 0 to 9, password protection, and custom encryption strength settings.
You install it globally on your machine using the Node Package Manager by running npm install -g zip-mcp. Once installed, add the zip-mcp command to the mcpServers configuration object of your preferred MCP client.
ZIP MCP Server is compatible with any tool implementing the Model Context Protocol. Supported clients explicitly documented include Claude Client, Cursor, and Raycast when configured with the MCP plugin.
Yes. Starting in version 1.0.6, the decompression tool validates each entry path. It rejects absolute paths, drive-letter specifications, and dot-dot segments to prevent malicious archives from writing files outside the intended destination directory.