vector_mcp
A server implementation for the Model Context Protocol (MCP) in Ruby.
claude mcp add --transport stdio sergiobayona-vector_mcp ruby -e require 'vector_mcp'; VectorMCP.new(name: 'VectorMCP', version: '0.0.0').run \ --env VECTOR_MCP_ENV="Placeholder for environment configuration"
How to use
VectorMCP is a Ruby-based MCP server framework that lets you expose tools, resources, prompts, and filesystem roots to clients that speak the MCP protocol. The server runs as a Ruby process using the vector_mcp gem, and provides multiple transports (notably stdio for CLI tools and SSE/HTTP for web integrations) out of the box. You define tooling by registering functions (tools), data sources (resources), and reusable prompts, then start the server to accept calls from LLMs or MCP-enabled clients. The framework emphasizes security through input validation, schema enforcement, and optional authentication/authorization hooks to restrict access to specific tools or resources.
You’ll interact with the server by creating a VectorMCP instance, registering your tools, resources, and prompts, and then calling run to start listening on the default stdio transport or on a specified transport/port if you enable HTTP/SSE. Typical workflows include exposing a calculator tool, reading configuration data as a resource, and supplying a code review or language-specific prompt template that the LLM can invoke.
To use the server effectively, install the gem, require it in your Ruby application, register the components you want to expose, and then run the server. Clients can connect and invoke the registered tools by sending MCP-compliant requests. The documentation also covers advanced security, filesystem roots, and real-world examples to help you tailor access and boundaries for production deployments.
How to install
Prerequisites:
- Ruby 2.7+ (preferably a recent Ruby version)
- RubyGems (bundler optional)
Install the vector_mcp gem:
gem install vector_mcp
Or, if you prefer bundler in a project:
bundle add vector_mcp
Create a simple server file (examples/vector_mcp_server.rb):
require 'vector_mcp'
server = VectorMCP.new(name: 'MyApp', version: '1.0.0')
server.register_tool(
name: 'greet',
description: 'Greets a user',
input_schema: {
type: 'object',
properties: { name: { type: 'string' } },
required: ['name']
}
) { |args| "Hello, #{args['name']}!" }
server.run # default stdio transport
Run the server:
ruby path/to/examples/vector_mcp_server.rb
If you want to customize transports (e.g., SSE/HTTP), modify server.run with transport options per the library’s API (e.g., server.run(transport: :sse, port: 8080)).
Upgrade notes: If you update the gem, adjust any breaking changes in your tool/resource definitions and verify your security and schema validations remain compatible with your MCP clients.
Additional notes
Notes and tips:
- This MCP server is implemented as a Ruby gem (VectorMCP). There is no dedicated npm/Node.js package for this server, so npm_package is null.
- Ensure your Ruby environment has bundler if you’re managing dependencies that way.
- Security is opt-in and relies on the gem’s authentication/authorization hooks; implement API keys or JWT strategies as needed for production.
- When exposing HTTP/SSE, consider enabling TLS in front of your server and validating inputs with the provided schemas to prevent injection or schema-bypass attacks.
- For filesystem roots, use the provided root registration methods to constrain tool operations within safe directories.
- Regularly run your test suite to ensure your registered tools handle edge cases and validate inputs as expected.
Related MCP Servers
jadx-ai
Plugin for JADX to integrate MCP server
sonarqube
Model Context Protocol (MCP) server for SonarQube
sandbox
A Model Context Protocol (MCP) server that enables LLMs to run ANY code safely in isolated Docker containers.
mcp_on_ruby
💎 A Ruby implementation of the Model Context Protocol
pluggedin-app
The Crossroads for AI Data Exchanges. A unified, self-hostable web interface for discovering, configuring, and managing Model Context Protocol (MCP) servers—bringing together AI tools, workspaces, prompts, and logs from multiple MCP sources (Claude, Cursor, etc.) under one roof.
turbomcp
A full featured, enterprise grade rust MCP SDK