MCP Stock Agent

Production-ready stock market analysis backend with automated CI pipeline and cloud deployment

🎯 Overview

This repository is a production-ready stock market analysis backend leveraging the Model Context Protocol (MCP) from Anthropic. The backend service enables AI agents to evaluate specific stocks through comprehensive analysis powered by real-time news aggregation from NewsAPI, Yahoo Finance, and Google.

Demo:

mcp_stock_agent_chatbot.mp4

Key Features:

• ✅ Automated CI Pipeline: GitLab CI for automated Docker image building and tagging
• ✅ Cloud-Native Design: Kubernetes orchestration with AWS EKS deployment
• ✅ MCP Protocol Implementation: Streamable HTTP-based communication between AI agent and tools
• ✅ Extensible Architecture: Modular MCP-based tool system allows seamless integration of additional data sources (e.g., history charts, images for deeper analysis)
• ✅ Modern Backend Stack: FastAPI with containerized architecture

---

🏗️ Architecture

┌─────────────┐      HTTP/MCP Protocol       ┌─────────────┐
│   FastAPI   │ ◄──────────────────────────► │ MCP Server  │
│   Client    │                              │  (Tools)    │
│  (GPT-4o)   │                              │             │
└─────────────┘                              └─────────────┘
                                                   │
                                                   ├─► NewsAPI
                                                   ├─► Yahoo Finance
                                                   └─► Google

Backend Components:

• Client Service: FastAPI application with GPT-4o integration for intelligent query processing
• MCP Server: Standardized HTTP service following Anthropic's Model Context Protocol for tool orchestration

---

🚀 Quick Start

Prerequisites

Python 3.10+
Docker & Kubernetes (for deployment)
OpenAI API Key
NewsAPI Key (free tier available at https://newsapi.org/)

Local Development

# Environment setup
python -m venv venv
source venv/bin/activate
pip install -r client/requirements.txt -r server/requirements.txt

# Configuration
cat > .env << EOF
OPENAI_KEY=your_openai_api_key
NEWS_API_KEY=your_newsapi_key
EOF

# Launch backend services
pip install python-dotenv
python -m dotenv -f .env run -- python server/mcp_server.py &
python -m dotenv -f .env run -- python client/service.py

Frontend Setup (Optional - AI-Assisted Demo UI)

cd ui
# Setup Tailwind
npm install -D tailwindcss postcss autoprefixer
npx tailwindcss init -p
# Install icon library
npm install lucide-react 
npm install
# Update apiUrl in src/App.tsx with your backend URL
npm run dev

---

📦 Production Deployment

CI Pipeline (GitLab)

Automated Docker image building and tagging on every commit via .gitlab-ci.yaml:

# Local test on built Docker images
make test

Kubernetes Deployment

1️⃣ Local Staging (Minikube)

# Create GitLab registry credentials
kubectl create secret docker-registry gitlab-registry-secret \
  --docker-server=registry.gitlab.com \
  --docker-username=<your-username> \
  --docker-password=<deploy-token> \
  --docker-email=<your-email>

# Deploy to staging
make staging_deploy

# Access backend service
# URL: http://$(minikube ip):30031

# Once staging service works, deploy release version 
make release
# URL: http://$(minikube ip):30032

2️⃣ AWS EKS Production

Setup AWS CLI & eksctl:

# Install AWS CLI
curl "https://awscli.amazonaws.com/awscli-exe-linux-x86_64.zip" -o "awscliv2.zip"
unzip awscliv2.zip && sudo ./aws/install
aws --version

# Configure credentials (requires IAM user with EKS permissions)
aws configure

# Install eksctl
curl --silent --location "https://github.com/weaveworks/eksctl/releases/latest/download/eksctl_$(uname -s)_amd64.tar.gz" | tar xz -C /tmp
sudo mv /tmp/eksctl /usr/local/bin
eksctl version

Deploy to EKS:

# Provision cluster (uses t3.micro free-tier nodes - adjust in aws-eks/eks-cluster.yaml for production)
eksctl create cluster -f aws-eks/eks-cluster.yaml

# Deploy application
make release_aws_eks

# Get service endpoint
kubectl get svc mcp-stock-agent -n agentic-release
# Access at: http://<EXTERNAL-IP>/docs

⚠️ Cost Management

# Always delete cluster after testing
eksctl delete cluster -f aws-eks/eks-cluster.yaml

# Verify deletion in AWS Console:
# - EC2 instances
# - EKS cluster
# - EBS volumes (if persistent storage was used)

---

🛠️ Technology Stack

Layer	Technology
Backend	FastAPI, Python
AI Integration	OpenAI GPT-4o
Protocol	Anthropic MCP (Model Context Protocol)
Containerization	Docker
Orchestration	Kubernetes
CI	GitLab CI
Cloud	AWS EKS

---

🔮 Future Enhancements

DevOps:

• Automated CD pipeline for direct EKS deployment
• Prometheus + Grafana monitoring stack
• Helm charts for templated deployments

Backend Features:

• Technical analysis APIs (price charts, indicators)
• Valuation metrics endpoints (P/E, P/B, DCF analysis)
• Stock screening API for undervalued opportunities
• Buy/sell recommendation engine

Performance:

• Explore open-source models (trade-off: reduced accuracy in complex tool-calling scenarios)
• Caching layer for frequently requested stock data
• Rate limiting and request optimization

---

📄 License

MIT License - See LICENSE file for details

---

👤 Contact

Cuong Van Dam - LinkedIn | Email