CAST MCP Servers
Overview
Ask your AI coding assistant about your applications’ architecture, transactions, dependencies and quality, and get answers based on CAST’s analysis of your code. An administrator installs the MCP Server once; each developer then adds it to their AI client.
MCP (Model Context Protocol) is the open standard that AI clients such as GitHub Copilot, Claude Code and Cursor use to call external tools.
The two CAST MCP Servers
| MCP Server | Ask about |
|---|---|
| CAST Imaging MCP | Application architecture, transactions, data flows, dependencies, quality insights and code objects |
| CAST Gatekeeper MCP | Compliance status, violations, technical debt, quality metrics and ISO 5055 |
If you use both products, connect to both: they share the same CAST Imaging URL and API key, and you add one entry for each to your AI client.
Get started
Configure your AI client - GitHub Copilot in Visual Studio Code, Claude Code or Claude Desktop - try example prompts and test the connection.
Install the MCP Servers, secure them with OAuth sign-in, and update and troubleshoot them.
Which LLMs give the best results with the CAST MCP Servers, and answers to security, data safety and deployment questions.