Magnolia MCP. Audit schemas and manage JCR structure via chat.

Magnolia CMS MCP Server connects your AI agent straight into an enterprise-grade headless content repository. This isn't just reading what hits the live site; you get full, direct control over Magnolia’s entire JCR structure and content management layer through natural conversation. You query the underlying data—the actual schema and nodes—not just the rendered website.

Inspecting the Content Structure

The server lets you audit component schemas using mg.get_template_schema. When you run it, you get a dump of all explicitly attached structured rules, letting you see exactly what fields a template or component expects to run correctly. Need to map out how content is delivered? You can fetch pure JSON mappings for any specific delivery endpoint path by calling mg.get_delivery_node, which identifies those bounded routing spaces within the JCR tree.

If you need to know all the components driving a page—like nested sections or branches—you run mg.get_delivery_children; this extracts properties and lists every single child node that contributes to the active branch nesting structure.

Querying and Discovering Data

You can poke around your repository using native JCR logic. By listing all active JCR workspaces with mg.list_jcr_workspaces, you know precisely which context domain you're operating in. Need to find specific bits of content? You run complex queries against property sets and specific content paths, giving you secure access to the underlying data.

For detailed auditing, you can use mg.query_delivery_nodes to retrieve explicit cloud logging data by tracing specific payload criteria within the repository.

Managing Content Nodes

This is where you make changes. You write new content using mg.create_cms_node, which generates and writes structured JSON payloads natively into a brand-new CMS node. If something already exists, you modify it directly with mg.patch_cms_node. This tool updates existing content nodes across global Web CRM boundaries, letting you patch draft document schemas in place without hassle.

Need to clean up? You can irreversibly delete an entire CMS node using mg.wipe_cms_node, which drops all associated live database bytes and content from the system. Furthermore, if a whole section needs to move—like copying a full article while keeping its structure intact—you use mg.copy_delivery_node; this copies a content node while maintaining the structural matching required for delivery logic.

Executing Operational Commands

The server lets you run platform commands that manage your entire workflow. You dispatch automated validation checks using mg.execute_workspace_command to route explicit system logic commands, which is crucial for publishing or activating content. By combining these tools, you can manage the complete lifecycle of content right from your agent’s conversation, handling everything from schema auditing and node creation to triggering final publication workflows.