Files.com

Files.com MCP enables your AI models to securely perform real-world file operations inside your Files.com environment. It exposes authentic, auditable actions to an LLM so you can upload, download, query folders and metadata, manage users, and automate workflows as if the model were part of your team.

How to use

Set up the MCP server in your LLM client to connect to Files.com via a local process (stdio) or a remote service (http) if provided. Once configured, your AI agent can perform actions such as transferring files, querying folders and metadata, creating or managing users, and triggering workflows like archival or sharing. Start with the simplest tools your agent needs and progressively enable more as you validate behavior. When using multiple tools, selectively enable only those required for the current task to reduce ambiguity and improve tool selection accuracy.

How to install

Prerequisites: you should have a runtime that supports the MCP client integration, and you will use uvx to run the Files.com MCP server adapter in stdio mode.

# Install uvx if you do not have it yet (the MCP examples assume uvx is available)
# On macOS/Linux:
# curl -sSf https://uv.pm/install | sh

# Alternatively, follow the uvx installation instructions for your platform

# Prepare your configuration (example JSON shown below) and save as files_com_mcp_config.json

# Run the MCP using uvx with your API key provided via environment variable or within the config

# Example start command (from the configuring snippet) mirrors the local stdio setup shown in the guide
"""
{
  "mcpServers": {
    "Files.com": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "files-com-mcp"
      ],
      "env": {
        "FILES_COM_API_KEY": "CHangeME"
      }
    }
  }
}
"""

Additional sections

Configuration details are provided in concrete code blocks below. You will typically run a local MCP process and point your LLM client to it, or connect to a remote MCP endpoint if you have one. Ensure you protect your API key and limit tool exposure to the minimum necessary for your agent’s tasks.

Configuration and examples

The process below demonstrates a standard stdio configuration using uvx to run the Files.com MCP server. This is the recommended approach when integrating with Claude or similar LLM clients that support stdio MCP wiring. You supply your Files.com API key and let the MCP client manage tool invocations securely.

{
  "mcpServers": {
    "Files.com": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "files-com-mcp"
      ],
      "env": {
        "FILES_COM_API_KEY": "CHangeME"
      }
    }
  }
}

Tools and capabilities

The MCP server enables a set of core file operations you can expose to your AI agent. Core capabilities include transferring files between cloud and on‑premises, querying folders and file metadata, creating and managing users, and automating workflows such as archival or sharing. Start with a focused subset of tools to minimize tool usage ambiguity and improve the model’s task success rate.

Available tools

transfer_files

Move files between cloud and on‑prem sources, enabling automated file transfers as part of LLM workflows.

query_metadata

Query folders and file metadata to inform decision making and provide context to the AI agent.

manage_users

Create or manage users within the Files.com environment as part of automation or provisioning tasks.

automate_workflows

Automate common file-related workflows such as archival, sharing, or scheduled tasks.