SCAST

SCAST MCP Server lets you run the SCAST static code analysis and visualization engine as an MCP server, enabling you to analyze codebases, generate UML diagrams and flowcharts, and retrieve structured insights through your AI client. It supports multiple programming languages and exposes analysis results as visual artifacts and keyword data that can power search and reasoning during conversations.

How to use

Connect your MCP client to the SCAST MCP server to analyze code, generate visual diagrams, and retrieve keyword-based summaries. Use the server to analyze a local codebase or workspace, then request diagrams such as UML, AST trees, and Mermaid flowcharts to understand code structure and functionality.

How to install

Prerequisites: Node.js and npm must be installed on your system.

Step by step commands to set up the MCP server locally:

{
  "mcpServers": {
    "scast":{
      "command":"node",
      "args":[
        "/YOUR_INSTALL_DIR/SCAST/mcp/index.js",
        "/YOUR_WORKSPACE/",
        "C:\\Users\\DKZ\\OTHER_ALLOWED_DIR\\",
      ]
    }
  }
}

Notes for using the MCP server

With the MCP server running, you can point your MCP client to the provided command and arguments. The configuration above uses a local, stdio-based server that executes the SCAST MCP entry point via Node.js.

Additional sections

Configuration and tools described here include analysis and retrieval capabilities intended for use within an MCP workflow. You can leverage the analysis tool to scan a folder and produce an AST tree, UML diagrams, and Mermaid flowcharts, then use the generated keywords to navigate the codebase or answer questions using a retrieval-augmented approach.

Available tools

scast_analysis

Analyzes code to generate an Abstract Syntax Tree (AST), visualizations (UML diagrams, AST diagrams, Mermaid flowcharts), and a keyword list with class/method names and explanations.

scast_retriever

Analyzes a code directory, builds an AST, and enables keyword-based searches to locate source definitions for explanation or retrieval, supporting a RAG-style workflow.