Egnyte MCP Server

The Egnyte MCP server is an official integration maintained by Egnyte that bridges AI agents to the Egnyte enterprise content platform. It exposes document search capabilities so clients like Claude Desktop, Cursor, ChatGPT, and other MCP-compatible tools can locate and reference files stored in an Egnyte domain. All access goes through Egnyte's public APIs and respects the organization's existing permission structure, so AI assistants only see content the authenticated user is authorized to view.

Egnyte ships two deployment options. The first is the open-source server in the egnyte/egnyte-mcp-server GitHub repository, a Python 3.11+ project that self-hosts and authenticates with an Egnyte API access token. The second is a fully managed Remote MCP Server hosted by Egnyte, which uses OAuth 2.0 and adds higher-level capabilities such as document summarization and AI Assistant Q&A on plans where it is available (Gen 4 Essential, Elite, and Ultimate).

The current open-source build is intentionally minimal: it exposes a single tool, search_for_document_by_name, for filename-based document lookup. Egnyte has indicated that future enhancements include hybrid search and document text extraction. For organizations standardizing on MCP for governed access to regulated content, this is the canonical Egnyte integration.

Tools

Prerequisites

• Python 3.11 or newer
• An Egnyte API access token (register at https://developers.egnyte.com/member/register)
• An active Egnyte domain with accessible files

Installation

Clone the repo and install the uv package manager and Egnyte SDK:

git clone https://github.com/egnyte/egnyte-mcp-server.git
cd egnyte-mcp-server

# Install uv (Mac/Linux)
curl -LsSf https://astral.sh/uv/install.sh | sh
# Windows PowerShell
# irm https://astral.sh/uv/install.ps1 | iex

uv pip install egnyte

Create a .env file in the project root:

DOMAIN=your-egnyte-domain.egnyte.com
ACCESS_TOKEN=your-access-token-here

Run the server locally to verify:

uv run server.py --python 3.11

Claude Desktop config

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "egnyte": {
      "command": "python3",
      "args": ["server.py"],
      "cwd": "/path/to/egnyte-mcp-server",
      "env": {
        "DOMAIN": "your-egnyte-domain.egnyte.com",
        "ACCESS_TOKEN": "your-access-token"
      }
    }
  }
}

Remote MCP option

Egnyte also offers a managed Remote MCP Server with OAuth 2.0 authentication on supported Gen 4 plans (Essential, Elite, Ultimate). Setup is done from the Egnyte admin console under Copilot settings. See the Egnyte helpdesk article "Egnyte MCP Server Overview" for the current remote endpoint and enablement steps.

• Have an AI assistant locate the latest version of a contract, SOW, or policy document by filename without leaving the chat
• Pull a specific report from Egnyte into a Claude or Cursor conversation as grounding context for summarization or analysis
• Let analysts ask a chatbot for "the Q3 financial deck" and surface the right file from a permission-controlled Egnyte folder
• Connect ChatGPT or VS Code Copilot to a regulated content store so agents only reference authorized enterprise files
• Build agent workflows (review, compare, summarize) that need governed file lookup as a first step

• "Find the latest MSA document for Acme Corp in Egnyte."
• "Search Egnyte for any file with 'Q3-forecast' in the name."
• "Locate the employee handbook PDF in our Egnyte domain."
• "Pull up the file called 'pricing-sheet-2026' from Egnyte."
• "Look up clinical trial protocol documents named 'CTP-' in Egnyte."

• Official, maintained by Egnyte under the egnyte GitHub org
• Honors existing Egnyte permissions, so AI agents cannot bypass access controls
• Simple, clean Python implementation that is easy to audit and self-host
• Managed Remote MCP option available for teams that do not want to operate the server

• Very limited tool surface: the open-source build exposes only filename search, with no full-text search, content read, upload, or sharing tools
• Document text extraction and hybrid search are listed as future work, not yet shipped
• Advanced features like AI Assistant Q&A and summarization are gated behind the managed Remote MCP Server and specific Egnyte plans

• Box MCP for similar enterprise content search on Box
• Google Drive MCP (reference server) for Drive-based file access
• SharePoint / Microsoft 365 MCP community servers for Microsoft content stores