Introducing the Hygraph MCP Server

In the rapidly evolving landscape of AI-assisted development, developers are looking for ways to integrate their content infrastructure directly into their AI workflows. The challenge has been clear: while AI assistants can write code and answer questions, they've been locked out of your actual content systems, forcing you to manually copy data back and forth or build custom integrations for every project.

We're excited to introduce the Hygraph MCP Server, now available in early access for all Hygraph projects. This powerful integration connects your Hygraph content directly to AI assistants through the Model Context Protocol (MCP), allowing you to manage, query, and publish content using natural language commands from tools such as Claude, Cursor, Windsurf, and VS Code.

#What is the Hygraph MCP Server?

The Hygraph MCP Server acts as a bridge between AI assistants and your Hygraph Content API, allowing large language models (LLMs) to interact with your project in a structured, secure, and standardized way.

Instead of switching between your IDE, the Hygraph Studio, and your API documentation, you can now work with your content entirely through natural conversation with your AI assistant. Ask questions about your schema, create content entries, update existing posts, or publish content, all without leaving your development environment.

This is fundamentally changing how developers and content teams interact with their CMS, making Hygraph a true AI-native content platform.

#Why MCP matters for content operations

The Model Context Protocol is an open standard that enables AI assistants to connect to external data sources and tools securely. By implementing MCP, Hygraph bridges the gap between your content API and your AI workspace, opening the door to entirely new developer workflows.

Consider these scenarios:

For Developers:

• Generate TypeScript types for all your content models without leaving your editor
• Fetch recent blog posts and automatically create markdown files with proper front matter

For Content Teams:

• Bulk update content that matches specific criteria ("Find all draft posts tagged 'urgent' and set their priority to high")
• Move content between environments with simple natural language commands
• Generate and publish content based on structured data or prompts

For Platform Teams:

• Build intelligent, compliant AI integrations without reinventing API logic for every project
• Perform bulk cleanups or migrations through conversational commands
• Automate content workflows that previously required custom scripts

The MCP Server reuses the same models and permissions as your Permanent Auth Token (PAT), ensuring that AI assistants respect your existing security and access control policies.

#How it works

Setting up the Hygraph MCP Server is straightforward and integrates directly into your existing workflow:

1. Authenticate & configure

The MCP Server authenticates using your Hygraph Permanent Auth Token (PAT). You retrieve your PAT and MCP Server API endpoint from your project settings, then configure your MCP client of choice. This ensures that all AI interactions respect your defined permissions, and only authorized models and operations are accessible.

2. Connect your AI Assistant

The Hygraph MCP Server supports most MCP-compatible tools. Whether you're using Claude Code from your terminal, Cursor or Windsurf IDEs for development, VS Code with Copilot, or Claude Desktop for general workflows, setup takes just a few minutes.

Each client handles HTTP servers slightly differently, but we've provided detailed configuration examples for all supported platforms in our documentation.

3. Start working with natural language

Once connected, your AI assistant can perform discovery, querying, and CRUD operations through natural conversation:

• "List all blog posts from Hygraph, then create individual markdown files for each."
• "Show me the structure of the Product model."
• "Find all posts with status DRAFT that have 'urgent' in the title and publish them."
• "Generate TypeScript types for all content models"

The MCP Server exposes tools that handle everything from schema discovery to publishing, turning your CMS into a conversational interface.

#Built to scale with you

The Hygraph MCP Server is designed to grow with your needs. Whether you're running a small side project or managing enterprise content operations across multiple environments, the same integration pattern scales seamlessly.

For teams managing multiple environments, you can configure separate MCP server instances for production, staging, and development, each with appropriately scoped tokens. This gives you the flexibility to test AI-driven workflows in staging before rolling them out to production.

#Getting started

The Hygraph MCP Server is available now in Early Access for all Hygraph projects, regardless of your plan tier.

Requirements:

• Node version >=22
• A Hygraph project with a Permanent Auth Token

Quick start:

1. Retrieve your PAT from Project Settings > Access > Permanent Auth Tokens
2. Copy your MCP Server API endpoint from Project Settings > Access > Endpoints
3. Configure your MCP client using our setup guides
4. Start interacting with your content through natural language

For detailed setup instructions specific to your IDE or AI assistant, refer to our comprehensive MCP Server documentation.

#What's next

The Hygraph MCP Server represents our first step toward making Hygraph a fully AI-native content platform. This Early Access release focuses on the core workflows that developers need most, but we're already planning enhancements based on your feedback.

As the Model Context Protocol ecosystem evolves and more AI assistants adopt MCP, your Hygraph content will be ready to work with them, requiring no additional integration work.