Unbundle OpenAPI MCP Server

Unbundle OpenAPI MCP Server is an MCP server that enables AI assistants to manipulate, split, and extract components from OpenAPI specifications using Redocly CLI. Built on Node.js and communicating over standard input/output (stdio), the server bridges local developer environments and AI clients such as Claude Desktop, Cursor, VS Code, and Cline. Software engineers, API designers, and technical writers use it to break down monolithic OpenAPI definition files into structured directories or create isolated specification files containing only specific endpoints and their referenced schemas. By invoking the underlying Redocly CLI via npx, the server provides programmatic tools to handle large API specs directly inside conversational coding workflows. This setup allows developers to target and bundle only the endpoints needed for specific microservices or context windows, avoiding manual schema editing and preventing context overload when feeding documentation to language models.

Category: Developer Tools & Code Intelligence

Tags: api-design, cli, openapi, redocly, swagger

Visit Unbundle OpenAPI MCP Server

How to install and configure Unbundle OpenAPI MCP Server

1. Ensure Node.js (v18 or v20 recommended) and npm are installed. 2. Configure your MCP client to launch the server via npx. 3. For Claude Desktop, edit claude_desktop_config.json and add: json { "mcpServers": { "unbundle_openapi": { "command": "npx", "args": ["@auto-browse/unbundle-openapi-mcp@latest"] } } } 4. For VS Code, Cline, or Cursor, add the configuration under mcp.servers in your settings.json or .vscode/mcp.json file. 5. Alternatively, install automatically for Claude Desktop using Smithery by running npx -y @smithery/cli install @auto-browse/unbundle_openapi_mcp --client claude. 6. Restart your MCP client application to load the tools.

What you can do with Unbundle OpenAPI MCP Server

• Splitting massive monolithic OpenAPI definition files into smaller modular component files using the split_openapi tool. - Extracting selected endpoints and their linked component schemas into a standalone bundled OpenAPI document via extract_openapi_endpoints. - Preparing reduced API documentation sub-specifications for specific client integrations or frontend development teams. - Minimizing prompt token usage in AI coding assistants by isolating only the relevant API paths needed for a task. - Automating OpenAPI structural reorganization within developer environments like Cursor or Claude Desktop without manual text editing.

Key facts

• https://github.com/auto-browse/unbundle_openapi_mcp
• Developer Tools & Code Intelligence, Files, Documents & PDFs
• api-design, cli, openapi, redocly, swagger

Part of MCP Servers

How do I install Unbundle OpenAPI MCP Server?

You can run the server directly via npx by configuring your client with the command npx and argument @auto-browse/unbundle-openapi-mcp@latest. Alternatively, Claude Desktop users can install it automatically using the Smithery CLI with the command npx -y @smithery/cli install @auto-browse/unbundle_openapi_mcp --client claude.

What tools does Unbundle OpenAPI MCP Server provide?

The server provides two primary tools: split_openapi and extract_openapi_endpoints. The split tool uses Redocly CLI to divide an OpenAPI specification into smaller modular files. The extract tool isolates specified paths, trims unneeded routes, and bundles the resulting endpoints and references into a new OpenAPI file.

Which MCP clients work with Unbundle OpenAPI MCP Server?

The server works with any MCP-compliant client that supports stdio communication. This includes Claude Desktop, Cursor, VS Code with MCP support, and Cline. Configuration is completed by adding the npx command details to the respective client configuration JSON file.

Does Unbundle OpenAPI MCP Server require Redocly CLI installed globally?

No, you do not need to install Redocly CLI globally beforehand. The server internally executes commands using npx @redocly/cli@latest to perform split and bundle actions, downloading or using the cached package as needed during execution.