This project demonstrates the use of a Bedrock Agents Action Group as a reusable tool across multiple Bedrock Agents. It now provides two key functionalities:
-
Stock Data Lookup: Retrieves a 1-month stock price history for a given ticker by leveraging the Yahoo Finance (yfinance) library.
-
Portfolio Optimization: Uses historical price data and a list of ticker symbols to compute an optimal portfolio. This function leverages portfolio optimization techniques from the pypfopt library, including the Efficient Frontier approach and discrete allocation based on latest prices.
The Portfolio Assistant Agent supervisor example demonstrates reusing this StockData tool. Note the implementation of the Lambda function currently ignores which Agent was used to call the Action Group and is not tightly coupled to any single Agent.
- AgentLambdaFunction: This AWS Lambda function implements the stock_data_lookup and portfolio_optimization functionalities. It uses a Python 3.12 container image with all dependencies packaged at build time.
- AgentLambdaRole: This is an AWS Identity and Access Management (IAM) role that grants the Lambda function the necessary permissions to execute.
- AgentAliasLambdaPermission and AgentLambdaPermission: These resources grant permissions for Amazon Bedrock Agents to invoke the Lambda function.
Deploy stock_data_stack.yaml
| Region | development.yaml |
|---|---|
| us-east-1 |  |
| us-west-2 | |
from src.utils.bedrock_agent import (
Agent,
region,
account_id,
)
import uuid
stock_data_tools_arn = f"arn:aws:lambda:{region}:{account_id}:function:stock_data_tools"
quantitative_analysis_agent = Agent.create(
name="quantitative_analysis_agent",
role="Financial Data Collector",
goal="Retrieve real-time and historic stock prices as well as optimizing a portfolio given tickers.",
instructions="""You are a Stock Data and Portfolio Optimization Specialist...
""",
tools=[
# Stock Data Lookup Tool
{
"code": stock_data_tools_arn,
"definition": {
"name": "stock_data_lookup",
"description": "Gets the 1-month stock price history for a given stock ticker, formatted as JSON.",
"parameters": {
"ticker": {"description": "The ticker to retrieve price history for", "type": "string", "required": True}
},
},
},
# Portfolio Optimization Tool
{
"code": stock_data_tools_arn,
"definition": {
"name": "portfolio_optimization",
"description": "Optimizes a stock portfolio given a list of tickers and historical prices from the stock_data_lookup function.",
"parameters": {
"tickers": {
"description": "A comma-separated list of stock tickers to include in the portfolio",
"type": "string",
"required": True
},
"prices": {
"description": "A JSON object with dates as keys and stock prices as values",
"type": "string",
"required": True
}
}
},
}
],
llm=LLM,
)
response1 = stock_data_agent.invoke( #invoking stock_data_lookup
input_text="What is the stock trend for AMZN?",
session_id: str = str(uuid.uuid1()),
enable_trace: bool = False,
)
print(response1)
response2 = stock_data_agent.invoke( #invoking portfolio_optimization
input_text= "Optimize a portfolio for the following: {\"2025-03-01\": {\"FAKE1\": 100.0, \"FAKE2\": 200.0, \"FAKE3\": 300.0}, \"2025-03-02\": {\"FAKE1\": 101.0, \"FAKE2\": 201.0, \"FAKE3\": 301.0}, \"2025-03-03\": {\"FAKE1\": 102.0, \"FAKE2\": 202.0, \"FAKE3\": 302.0}}",
session_id: str = str(uuid.uuid1()),
enable_trace: bool = False,
)
print(response2)
response3 = stock_data_agent.invoke( #invoking both tools
input_text="Optimize a portfolio with AMZN, GOOG, and MSFT",
session_id: str = str(uuid.uuid1()),
enable_trace: bool = False,
)
print(response3)- Open the CloudFormation console.
- Select the stack
StockDataLookupyou created, then click Delete. Wait for the stack to be deleted.
See CONTRIBUTING for more information.
This project is licensed under the Apache-2.0 License.s