gepa-adk 0.3.4

Evolutionary optimization for Google ADK agents

gepa-adk

Evolutionary optimization for Google ADK agents.

What is this?

gepa-adk evolves AI agent instructions automatically. Give it an agent and training examples, and it finds better prompts through iterative improvement.

Installation

pip install gepa-adk

export GOOGLE_API_KEY=your-api-key  # or other ADK-supported model

Quick Start

Evolve a greeting agent to produce formal, Dickens-style greetings:

import asyncio
from google.adk.agents import LlmAgent
from gepa_adk import evolve, EvolutionConfig, SimpleCriticOutput

agent = LlmAgent(
    name="greeter",
    model="gemini-2.5-flash",
    instruction="Greet the user appropriately.",
)

critic = LlmAgent(
    name="critic",
    model="gemini-2.5-flash",
    instruction="Score for formal, Dickens-style greetings. 0.0-1.0.",
    output_schema=SimpleCriticOutput,
)

trainset = [
    {"input": "I am His Majesty, the King."},
    {"input": "I am your mother."},
    {"input": "I am a close friend."},
]

config = EvolutionConfig(
    max_iterations=5,
    patience=1,
    reflection_model="gemini-2.5-flash",  # Model for generating improvements
)
result = asyncio.run(evolve(agent, trainset, critic=critic, config=config))
print(f"Score: {result.original_score:.2f} -> {result.final_score:.2f}")
print(result.evolved_components["instruction"])

Examples

• basic_evolution.py — Single agent with critic
• critic_agent.py — Custom critic for stories
• multi_agent.py — Multi-agent evolution

Documentation

Getting Started · Guides · API Reference

Credits

Based on GEPA (source). Built on Google ADK (source).

License

Apache 2.0