Dropbox-like web file-manager supporting many back-ends (SFTP, FTP/S, WebDAV, S3, SMB, Git, LDAP, MySQL, etc.). Comes with plugin system, media viewers/transcoders and shares.
https://github.com/mickael-kerjean/filestashStop juggling multiple file management tools. This MCP server puts Filestash's universal file management capabilities directly into your AI workflow, giving you seamless access to files across SFTP, S3, Git, WebDAV, and a dozen other storage systems through a single interface.
You're managing deployment scripts on SFTP, documentation in S3, configuration files in Git repos, and client deliverables on WebDAV shares. Each system needs different tools, authentication methods, and workflows. Your AI assistant can't help because it can't reach these scattered files.
This MCP server exposes Filestash's proven multi-backend architecture (11k+ GitHub stars) to your AI models. Configure your storage connections once, then work with files anywhere as if they're local.
Supported backends out of the box:
Deployment automation: Your AI can pull configs from Git, upload to SFTP, and verify deployment by checking S3 logs - all in one conversation.
Documentation sync: Edit markdown files locally, have your AI push them to multiple storage backends, then generate deployment summaries from server logs.
Client project handoffs: AI can package files from development Git repos, upload to client WebDAV shares, and create organized folder structures with proper permissions.
Database documentation: Extract schema information from MySQL, generate documentation, and automatically distribute to team SMB shares and project S3 buckets.
Filestash handles the authentication complexity. Set up your storage connections through the web interface once - the MCP server inherits all configured backends automatically. No credential juggling, no per-backend API learning curves.
The server maintains persistent sessions across all your storage systems, so your AI workflows don't break when tokens expire or connections reset.
Drop this into your existing MCP setup:
docker run -d -p 8334:8334 machines/filestash
# Configure your backends at http://localhost:8334
# Add MCP server to your client configuration
Your AI models immediately gain file operations across every configured backend. List directories, read files, write content, create folders - the same commands work whether you're hitting local storage or remote S3 buckets.
Plugin extensibility: Need custom storage integration? Filestash's Go plugin system lets you add new backends without touching core code. Your MCP server automatically exposes new capabilities as you add them.
This isn't about replacing your existing tools - it's about giving your AI assistant the same universal file access you need for modern development workflows. Set it up once, then focus on building instead of managing storage connections.