Getting started with CAST Imaging MCP Server


An MCP (Model Context Protocol) server is a standardized way for AI applications to connect to external data sources and tools. MCP servers act as intermediaries that expose resources (like files, databases, APIs) and tools (like functions, scripts) to AI models through a standardized interface. This creates a secure bridge between the AI and external systems.

In the context of CAST Imaging, an MCP server is required if you want to interface AI clients such as Github Copilot and Claude Desktop with CAST Imaging results via its API.

Starting release 3.1.0-funcrel, the installation media delivers an MCP Aggregator Gateway and three MCP servers behind it. Your MCP-aware client continues to connect to CAST Imaging, which redirects MCP requests to the MCP Aggregator Gateway; the gateway then routes each request to the MCP server able to answer it, presenting the tools of all enabled MCP servers - structural, semantic search and Text2Cypher - as a single tool surface. Compared with MCP Server 3.0.x:

  • the configuration file changes - a single .env file on Linux via Docker, replacing config/app.config
  • the port CAST Imaging redirects MCP requests to changes from 8282 to 8285
  • a Neo4j instance is required by the semantic search and Text2Cypher MCP servers
  • the tool surface profile is now set per MCP server

See the installation and configuration instructions below.

Architecture

How the MCP Aggregator Gateway aggregates the structural, semantic search and Text2Cypher MCP servers behind a single endpoint: components, tool surface, ports, HTTP endpoints and security posture.

Microsoft Windows

This option is specifically for a direct installation on your Microsoft Windows machine.

Linux via Docker

/ This option is specifically for a direct install on Docker running on a Linux operating system. All components are provided as Docker images.

Meta Tool mode

The meta-tool architecture allows an AI aware client to work with CAST Imaging through a small set of general-purpose tools rather than loading (around 50) individual tools in advance. The tool surface profile is set per MCP server.

FAQ

Answers to a series of security and data safety related frequently asked questions about CAST Imaging’s MCP (Model Context Protocol) server.


MCP Server 3.0.x

Installation instructions for MCP Server 3.0.x and earlier. Installation on Linux via Podman and in an air-gapped environment is documented for 3.0.x only and has not yet been revised for 3.1.0.