Bedrock Stock Agent

An AI agent solution on AWS that hosts a FastAPI endpoint for querying real-time and historical stock prices, powered by LangGraph and Bedrock Claude models with Cognito authentication.

Features

• ReAct Agent: LangGraph-based agent using Claude 3.5 Haiku via AWS Bedrock
• Real-time Stock Prices: Query current stock prices via yfinance
• Historical Data: Retrieve historical stock price trends
• Document RAG: Search Amazon financial documents (Annual Report, Earnings Releases)
• Streaming Responses: Stream agent reasoning and tool calls via Server-Sent Events
• Cognito Authentication: JWT-based authentication via AWS Cognito
• AWS Deployment: Dockerized application deployed via Bedrock Agentcore

Architecture

Prerequisites

• AWS Account with permissions to:
  • Access Bedrock models
  • Create Cognito user pools
  • Push images to ECR
  • Create Agentcore runtimes
• Terraform >= 1.0
• Python 3.11+
• Docker

Quick Start

1. Clone and Setup

git clone <repository>
cd bedrock-stock-agent

# Create virtual environment
python -m venv venv
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Copy and configure environment
cp .env.example .env
# Edit .env with your actual values

2. Local Development

Run FastAPI Locally

export ENVIRONMENT=dev
export SKIP_AUTH=true

uvicorn src.api.main:app --reload --host 0.0.0.0 --port 8000

# Test health
curl http://localhost:8000/ping

# Test invoke (no auth in dev mode)
curl -X POST http://localhost:8000/invocations \
  -H "Content-Type: application/json" \
  -d '{"input": {"prompt": "What is the stock price for Amazon right now?"}}'

3. Deploy to AWS

Infrastructure (Terraform)

Terraform manages the supporting infrastructure: Cognito, S3, and IAM.

cd terraform

terraform init
terraform plan
terraform apply
terraform output -json

4. Test the Deployment

Open test_agent.ipynb and run the cells. The notebook reads credentials from .env and tests:

1. Cognito authentication
2. Health check
3. Agent invocation with sample queries

API Endpoints

Method	Path	Description
GET	/ping	Health check
POST	/invocations	Bedrock Agentcore standard endpoint

Project Structure

bedrock-stock-agent/
├── src/
│   ├── agent/
│   │   ├── graph.py              # LangGraph ReAct agent
│   │   └── tools.py              # Stock and retrieval tools
│   ├── api/
│   │   ├── main.py               # FastAPI application
│   │   └── auth.py               # Cognito authentication
│   └── knowledge/
│       ├── loader.py             # Document loading
│       └── retriever.py          # FAISS vector store
├── terraform/
│   ├── main.tf                   # Provider and backend
│   ├── cognito.tf                # Cognito resources
│   ├── iam.tf                    # IAM roles and policies
│   ├── s3.tf                     # S3 document storage
│   ├── agentcore.tf              # ECR image output
│   ├── variables.tf              # Input variables
│   └── outputs.tf                # Output values
├── data/                          # Amazon financial documents
├── image/                         # Langfuse test
├── test_agent.ipynb               # Test notebook
├── Dockerfile                     # Container definition
├── requirements.txt               # Python dependencies
└── .env.example                   # Environment variable template

Configuration

Environment Variables

Copy .env.example to .env and fill in values:

# AWS Configuration
AWS_REGION=us-east-1
ENVIRONMENT=dev

# Cognito Authentication
COGNITO_CLIENT_ID=<your-cognito-client-id>
COGNITO_USERNAME=<your-cognito-username>
COGNITO_PASSWORD=<your-cognito-password>

# Bedrock Configuration
USE_BEDROCK_EMBEDDINGS=false
BEDROCK_MODEL_ID=us.anthropic.claude-3-5-haiku-20241022-v1:0

# AgentCore Configuration
AGENTCORE_RUNTIME_ARN=<your-agentcore-runtime-arn>

# Langfuse Observability
LANGFUSE_SECRET_KEY=<your-langfuse-secret-key>
LANGFUSE_PUBLIC_KEY=<your-langfuse-public-key>
LANGFUSE_HOST=https://us.cloud.langfuse.com

For local development, set SKIP_AUTH=true to bypass Cognito authentication.

Troubleshooting

Agent not responding

• Check AWS credentials: aws sts get-caller-identity
• Verify Bedrock access: aws bedrock list-foundation-models --region us-east-1

Documents not loading

• Verify files in data/ directory
• Check S3 bucket permissions
• Rebuild vector store: rm -rf .vector_store

Cognito authentication failing

• Verify user exists: aws cognito-idp admin-get-user --user-pool-id <ID> --username <user>
• Check token validity (tokens expire after 1 hour)
• Verify client ID matches .env

Cleanup

cd terraform
terraform destroy