MCP Integration

Model Context Protocol server with 7 production tools for intelligent portfolio interactions. Enables semantic search, interview practice, and dynamic profile queries through a FastAPI backend.

System Architecture

FastAPI Server

Python MCP server on port 3000

7 production tools
HTTP + stdio protocols
CORS enabled

Next.js Frontend

Portfolio app on port 3001

Consumes MCP tools
Real-time chat interface
Interactive demos

RAG Pipeline

Semantic search system

Upstash Vector DB
Groq LLaMA 3.3
384-dim embeddings

Available MCP Tools

get_portfolio_summary

Get a concise summary of the portfolio owner's role, summary, and skills

Portfolio

list_star_items

List all STAR methodology items (professional experiences/projects)

Portfolio

get_star_item

Get details of a specific STAR item by ID

Portfolio

Parameters:

item_id: string*

search_skills

Search for STAR items by skill/technology

Portfolio

Parameters:

skill: string*

get_interview_questions

Get common interview questions and prepared answers

Interview

Parameters:

limit: number

practice_interview

Get a random interview question for practice

Interview

semantic_search

Perform RAG search across all portfolio content

RAG

Parameters:

query: string*top_k: number

Integration Examples

Portfolio Chat Integration

Main chat interface uses semantic_search tool to find relevant portfolio content

Profile Data Page Integration

Profile page displays STAR items fetched from MCP tools

Interview Training Integration

Practice interview questions with prepared answers

Testing & Development

Multiple ways to test and interact with MCP tools:

Command Line

python scripts/test_mcp_direct.py

Direct Python testing of all tools

HTTP API

curl localhost:3000/tools/*

REST API endpoints for each tool

Chat Interface

localhost:3001 →

Natural language queries with RAG

Interactive Testing

/testing →

Web UI for manual tool execution

Claude Desktop

MCP stdio protocol

Native integration with Claude Desktop app