Update ADK doc according to issue #1292 - 4 - Prompt Optimization

docs/agents/prompt-optimization.md

@@ -0,0 +1,55 @@
+# Prompt Optimization
+
+The Simple Prompt Optimizer is a feature that helps you iteratively improve your agent's system prompts using a large language model (LLM).
+
+## How it works
+
+The optimizer uses an optimizer LLM (default `gemini-1.5-flash`) to critique and rewrite the prompt based on evaluation scores. It works as follows:
+
+1.  **Evaluates the initial prompt**: The optimizer first evaluates the agent with the initial prompt to get a baseline score.
+2.  **Generates a new prompt**: It then uses the optimizer LLM to generate a new, improved prompt.
+3.  **Evaluates the new prompt**: The optimizer evaluates the agent with the new prompt.
+4.  **Compares and updates**: If the new prompt results in a better score, it replaces the old prompt with the new one.
+5.  **Repeats**: The optimizer repeats this process for a set number of iterations.
+
+## Usage
+
+To use the Simple Prompt Optimizer, you need to:
+
+1.  Define an `Agent` and a `Sampler` (for evaluation).
+2.  Configure the `SimplePromptOptimizerConfig`.
+3.  Run the optimizer's `optimize` method.
+
+```python
+from google.adk.agents import Agent
+from google.adk.optimization import SimplePromptOptimizer, SimplePromptOptimizerConfig
+from google.adk.optimization import Sampler
+
+# 1. Define an Agent and a Sampler
+my_agent = Agent(
+    instruction="You are a helpful assistant.",
+    ...
+)
+
+my_sampler = Sampler(...)
+
+# 2. Configure SimplePromptOptimizerConfig
+optimizer_config = SimplePromptOptimizerConfig(
+    num_iterations=10,
+    batch_size=5,
+)
+
+# 3. Run the optimizer
+optimizer = SimplePromptOptimizer(config=optimizer_config)
+optimization_result = await optimizer.optimize(
+    initial_agent=my_agent,
+    sampler=my_sampler,
+)
+
+best_agent = optimization_result.optimized_agents[0].optimized_agent
+```
+
+## Key Classes
+
+*   **`SimplePromptOptimizer`**: The main class that runs the optimization process.
+*   **`SimplePromptOptimizerConfig`**: A data class for configuring the optimizer, including the number of iterations and the batch size for evaluation.