MCP Workspace Server

You deploy and run an All-in-One MCP Server that gives your AI agent complete capabilities like file operations, code execution, web deployment, data processing, and image generation in a single, secure workspace. Configure once, then manage all features from one place, with multi-tenant isolation and safe sandboxed environments.

How to use

To use this MCP Server with an AI platform, connect via SSE or a standard HTTP API and start issuing tasks that span file handling, code execution, web app deployment, data processing, and image generation. The server provides a virtual, isolated workspace for each user session, so your AI agent can read, write, run code, deploy frontends, process Excel data, and render diagrams without exposing the underlying file system.

How to install

Prerequisites: you need Docker installed on your machine for recommended deployment. You may also run locally with Node/Python environments if you choose to set up manually, but Docker is strongly advised for reliability.

# Clone the project repository
# Replace <repository-url> with the actual repository URL
git clone <repository-url>
cd mcp-filesystem

# First-time deployment: build images and start containers
docker-compose up -d --build

# If you cannot access the image registry, disable BuildKit as needed
export DOCKER_BUILDKIT=0

# After code updates, restart to apply changes
git pull && docker-compose restart

# View logs in real time
docker-compose logs -f

# Rebuild only if dependencies change
docker-compose up -d --build

Additional setup and configuration

One MCP Server handles all capabilities. You can enable or disable features via configuration options and environment variables. The server supports multi-tenant sessions with isolated workspaces and a virtual path layer that keeps the real server directory hidden from AI agents.

Configuration and deployment notes

CI/CD and production readiness are supported through a production-grade deployment workflow. The server exposes an admin interface for monitoring and a Web UI to inspect the file tree and preview text, Markdown, and CSV files in user workspaces.

Security and isolation best practices

Maintain strict multi-tenant isolation by issuing unique user and session identifiers per request. Use HTTPS in production, enforce request throttling, and keep a clear boundary around allowed directories. Regularly rotate admin credentials and restrict access to admin endpoints.

Usage examples

Web development workflow: create a frontend project, deploy it with a single action, and access it on a dedicated subdomain. Data analysis: read Excel files, process data, generate charts, and publish a report dashboard. Code development: write Python scripts, run tests, fix bugs, and deploy an API service.

Available tools

fs_write

Create or overwrite files (HTML/CSS/JS) with automatic format detection for frontend projects.

preview_frontend

One-click deployment of frontend projects with automatic independent subdomains and HTTPS when configured.

exec

Execute Python or Node.js code in a sandbox with real-time debugging support.

generate_image

Generate charts, mermaid diagrams, data visualizations, and HTML-rendered images.

fs_read

Read files, support for batches, Excel, and range-based access.

fs_ops

General filesystem operations such as list, mkdir, move, info, and delete.

fs_replace

Edit files precisely using a diff-like mechanism with search/replace.

fs_search

Search files by name (glob) or by content (regex) with optional context.

excel_edit

Edit Excel files by updating cells, applying formats, and batch changes.

list_excel_templates

List available Excel templates for creation or templating.

create_excel_from_template

Create Excel files from predefined templates.

kb_search

Glob-style search across enterprise knowledge bases.

kb_read

Read knowledge base content (Markdown output).

crawl_url

Fetch webpages and return content as Markdown.

web_search

Internet search capability to gather information.