04-llm-agents

04. LLM Agents

Building intelligent LLM-powered agents with Google ADK.

Blog Post: https://arjunprabhulal.com/adk-llm-agents/

Table of Contents

1. What is an LLM Agent?
2. Prerequisites
3. Setup Steps
4. Agent Configuration
5. Running the Agent
6. Next Steps

What is an LLM Agent?

An LLM Agent uses a Large Language Model as its reasoning engine to:

• Understand user intent
• Plan and execute actions
• Generate contextual responses
• Maintain conversation state

Prerequisites

• Python 3.10+
• Gemini API key from AI Studio

Setup Steps

1. Navigate to this module:

cd 04-llm-agents

2. Create and activate a virtual environment:

python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

3. Install dependencies:

pip install -r ../requirements.txt

4. Set up environment variables in financial_analyst/.env:

GOOGLE_API_KEY=your-api-key-here

Agent Configuration

Key parameters when defining an LLM Agent:

Parameter	Description	Example
model	The LLM model to use	gemini-2.5-flash
name	Unique agent identifier	fin_analyst
instruction	System prompt defining behavior	Role, expertise, guidelines
description	What the agent does	Used for routing in multi-agent systems

Running the Agent

Using ADK Web

adk web

Open http://127.0.0.1:8000 and select financial_analyst.

Test Queries:

• "What is a PE ratio?"
• "Explain the difference between growth and value stocks"
• "What are the key metrics in a 10-K filing?"

Using ADK CLI

adk run financial_analyst

Next Steps

Continue to 05. Workflow Agents