AWS Strands旅行智能体

Prerequisites

- AWS Bedrock with foundational text LLMs enabled. - AWS Agentcore - Verify that AgentCore is available in your tenant and region for deployment. Refer to Agentcore FAQ - AWS CLI - Install AWS CLI by following the installation instructions - AWS credentials - This value can be obtained by running AWS CLI aws sts get-session-token in the Terminal. For more options please refer to AWS instructions on CLI auth. - Use the two values from the output:

    "Credentials": {
        "AccessKeyId": "<VALUE>",
        "SecretAccessKey": "<VALUE>"
    }
    

• Okahu cloud account
• Sign up on Okahu portal
• Okahu API key
• Go to https://portal.okahu.co/settings
• Generate API key and copy it

Setup demo environment

• Start a command line shell
• Create a Python virtual environment
• python -m venv .venv
• Activate the virtual environment
• On macOS/Linux: source .venv/bin/activate
• On Windows (Git Bash): source .venv/bin/activate
• On Windows (Command Prompt): .venv\Scripts\activate.bat
• On Windows (PowerShell): .venv\Scripts\Activate.ps1
• Install dependencies in your python environment
• pip install -r requirements.txt
• Run Okahu demo setup tool. This will deploy demo agent to AWS Agentcore and setup Okahu tenant for consuming traces from that deployment.

•  --key <OKAHU-API-KEY>

• Verify that you see Demo setup completed successfully at the end

Try out tests framework example for the travel agent app

• Start a command line shell
• Create a Python virtual environment
• python -m venv .venv
• Activate the virtual environment
• On macOS/Linux: source .venv/bin/activate
• On Windows (Git Bash): source .venv/bin/activate
• On Windows (Command Prompt): .venv\Scripts\activate.bat
• On Windows (PowerShell): .venv\Scripts\Activate.ps1
• Source python env
• Install python dependencies
• pip install -r requirements.txt
• Run pytest
• pytest -vv test/test_travel_agent.py

Test locally using Agentcore CLI tool

• Start command shell and source python env
• Run Agentcore CLI command
• agentcore invoke '{"prompt": "<prompt>"}'