MCP Ecosystem (Model Context Protocol)

The MCP Ecosystem is a collection of standardized servers I engineered to give Artificial Intelligence Agents (LLMs) “hands and eyes” in the real world, fully compliant with Anthropic’s emerging Model Context Protocol standard.

Core Components

• BrowserMCP: A dual-engine server (Static + Playwright) enabling AI to autonomously navigate the web, interact with DOM elements, handle modern SPAs, and execute workflows.
• ScrapeMCP: Specialized data extraction server capable of parsing sitemaps, extracting tables, and recursively crawling sites to output structured semantic Markdown or JSON.
• RouteMCP: Intelligent AI router that classifies incoming tasks and delegates them to the most cost/latency efficient model (Gemini, Groq, Cerebras).
• DocMCP: Deep PDF manipulation tool allowing agents to read, merge, split, and generate analytical reports from office documents.

Architecture

Developed across TypeScript / Node.js (for asynchronous Playwright DOM manipulation) and Python (for document parsing and routing logic). These tools represent the state-of-the-art in Tool Calling and agentic autonomy.