Merge pull request #491 from ddobrin:main

PiperOrigin-RevId: 826517837

Author: copybara-github

.gitignore

@@ -29,3 +29,7 @@ out/
 # OS-specific junk
 .DS_Store
 Thumbs.db
+
+# Local documentation and plans
+docs/
+plans/

contrib/spring-ai/DOCUMENT-GEMINI.md

@@ -0,0 +1,86 @@
+# Documentation for the ADK Spring AI Library
+
+## 📖 Overview
+The `google-adk-spring-ai` library provides an integration layer between the Google Agent Development Kit (ADK) and the Spring AI project. It allows developers to use Spring AI's `ChatModel`, `StreamingChatModel`, and `EmbeddingModel` as `BaseLlm` and `Embedding` implementations within the ADK framework.
+
+The library handles the conversion between ADK's request/response formats and Spring AI's prompt/chat response formats. It also includes auto-configuration to automatically expose Spring AI models as ADK `SpringAI` and `SpringAIEmbedding` beans in a Spring Boot application.
+
+## 🛠️ Building
+To include this library in your project, use the following Maven coordinates:
+
+```xml
+<dependency>
+    <groupId>com.google.adk</groupId>
+    <artifactId>google-adk-spring-ai</artifactId>
+    <version>0.3.1-SNAPSHOT</version>
+</dependency>
+```
+
+You will also need to include a dependency for the specific Spring AI model you want to use, for example:
+```xml
+<dependency>
+    <groupId>org.springframework.ai</groupId>
+    <artifactId>spring-ai-openai</artifactId>
+</dependency>
+```
+
+## 🚀 Usage
+The primary way to use this library is through Spring Boot auto-configuration. By including the `google-adk-spring-ai` dependency and a Spring AI model dependency (e.g., `spring-ai-openai`), the library will automatically create a `SpringAI` bean. This bean can then be injected and used as a `BaseLlm` in the ADK.
+
+**Example `application.properties`:**
+```properties
+# OpenAI configuration
+spring.ai.openai.api-key=${OPENAI_API_KEY}
+spring.ai.openai.chat.options.model=gpt-4o-mini
+spring.ai.openai.chat.options.temperature=0.7
+
+# ADK Spring AI configuration
+adk.spring-ai.model=gpt-4o-mini
+adk.spring-ai.validation.enabled=true
+```
+
+**Example usage in a Spring service:**
+```java
+import com.google.adk.models.BaseLlm;
+import com.google.adk.models.LlmRequest;
+import com.google.adk.models.LlmResponse;
+import org.springframework.beans.factory.annotation.Autowired;
+import org.springframework.stereotype.Service;
+import reactor.core.publisher.Mono;
+
+@Service
+public class MyAgentService {
+
+    private final BaseLlm llm;
+
+    @Autowired
+    public MyAgentService(BaseLlm llm) {
+        this.llm = llm;
+    }
+
+    public Mono<String> generateResponse(String prompt) {
+        LlmRequest request = LlmRequest.builder()
+            .addText(prompt)
+            .build();
+        return Mono.from(llm.generateContent(request))
+            .map(llmResponse -> llmResponse.content().get().parts().get(0).text().get());
+    }
+}
+```
+
+## 📚 API Reference
+### Key Classes
+- **`SpringAI`**: The main class that wraps a Spring AI `ChatModel` and/or `StreamingChatModel` and implements the ADK `BaseLlm` interface.
+  - **Methods**:
+    - `generateContent(LlmRequest llmRequest, boolean stream)`: Generates content, either streaming or non-streaming, by calling the underlying Spring AI model. It converts the ADK `LlmRequest` to a Spring AI `Prompt` and the `ChatResponse` back to an ADK `LlmResponse`.
+
+- **`SpringAIEmbedding`**: Wraps a Spring AI `EmbeddingModel` to be used for generating embeddings within the ADK framework.
+
+- **`SpringAIAutoConfiguration`**: The Spring Boot auto-configuration class that automatically discovers and configures `SpringAI` and `SpringAIEmbedding` beans based on the `ChatModel`, `StreamingChatModel`, and `EmbeddingModel` beans present in the application context.
+
+- **`SpringAIProperties`**: A configuration properties class (`@ConfigurationProperties("adk.spring-ai")`) that allows for customization of the Spring AI integration.
+  - **Properties**:
+    - `model`: The model name to use.
+    - `validation.enabled`: Whether to enable configuration validation.
+    - `validation.fail-fast`: Whether to fail fast on validation errors.
+    - `observability.enabled`: Whether to enable observability features.