Powered by Zoomin Software. For more details please contactZoomin

MarkLogic MCP Server

Table of Contents

Introduction

Save PDF
Save selected topicSave selected topic and subtopicsSave all topics
Share
Share to emailCopy topic URL
Print
  • Last Updated: September 9, 2026
  • 3 minute read
    • Documentation

The MarkLogic Model Context Protocol (MCP) Server enables AI assistants to search, retrieve, and augment content from existing MarkLogic Server databases. Organizations that use AI assistants such as GitHub Copilot or Claude can integrate MarkLogic Server as a data source, so the AI can access and reason about organizational data during conversations and code generation tasks.

System overview

MarkLogic MCP Server operates as a bridge between AI assistants and MarkLogic Server databases to enable seamless integration of MarkLogic Server data into AI-assisted workflows. The system comprises three primary components that work in concert to deliver search and retrieval capabilities:

  • Your MCP client: An AI assistant such as GitHub Copilot or Claude. The client discovers and invokes tools through the Model Context Protocol (MCP), a standardized interface that exposes tool capabilities to AI systems. The client does not interact directly with MarkLogic Server; instead, it sends requests to the MarkLogic MCP Server.

  • MarkLogic MCP Server: A containerized service that handles communication between MCP clients and MarkLogic Server. MarkLogic MCP Server receives tool calls from an MCP client, translates them into MarkLogic Retrieval API requests, forwards those requests to MarkLogic Server, and returns results to the client. MarkLogic MCP Server handles request routing, authentication token management, and response formatting but does not itself perform search, ranking, or data retrieval operations.

  • MarkLogic Retrieval API (MRA): A set of REST endpoints deployed within MarkLogic Server that performs the actual search, filtering, retrieval, and augmentation of data. MarkLogic Retrieval API operates on existing MarkLogic Server databases and indexes, making use of MarkLogic Server's native search, vector, and relational capabilities. MarkLogic Retrieval API is installed into a MarkLogic REST app server and exposed through three primary endpoints:

    • /v1/retrieve for hybrid search
    • /v1/retrieve/definition for discovery of available filters and capabilities
    • /v1/retrieve/augment for full document retrieval by URI

Requests flow through the system like this:

  1. When an AI assistant receives a query from a user, it invokes a tool through the MCP protocol.

  2. MarkLogic MCP Server receives this request, authenticates it, and forwards it to the MarkLogic Retrieval API.

  3. MarkLogic Retrieval API processes the request using MarkLogic Server's search infrastructure, applies filters and constraints, and returns results in a standardized format.

  4. The MCP Server reformats these results and returns them to the MCP client, which uses the data to augment its response to the user.

For details on request flow and system boundaries, see Architecture. For details on request validation and security, see Security model.

Guide overview

This guide explains how the MarkLogic MCP Server works and provides instructions for installing and configuring the system:

  • Quick start: For users who are ready to deploy immediately, this section provides a step-by-step walkthrough covering the installation of the MarkLogic Retrieval API, configuration of the MCP server, and initial verification. This section is intended for environments with existing MarkLogic Server deployments and assumes basic familiarity with MarkLogic Server and Docker.

  • Concepts: For users seeking to understand the system architecture before deployment, this section provides detailed explanations of how the MCP Server and Retrieval API are designed, how they interact with MarkLogic Server, and how they expose capabilities to AI assistants.

  • Operations: For users who have initially deployed, this section provides task-specific guidance for configuring MCP clients, hardening the system for production, and troubleshooting deployment issues.

  • Reference: For users who need comprehensive technical details, this section provides these details for the MarkLogic Retrieval API endpoints, the MCP Server configuration schema, and runtime environment variables.

Support

GitHub Issues lists any current MCP Server issues.

Contributions

CONTRIBUTING.md details how to contribute to the MarkLogic MCP Server project.

License

This project is licensed under the Apache License 2.0. See the LICENSE file for details.

Alert