### Build Project

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Command to clean and install the project using Maven.

```bash
./mvnw clean install
```

--------------------------------

### Example: Chat History Before Compaction

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Shows a sample chat history with 80 messages before compaction is applied.

```text
1. User: What's the capital of France?
2. Assistant: The capital of France is Paris.
3. User: Tell me about its history.
4. Assistant: Paris has a rich history dating back to...
...
79. User: What about modern Paris?
80. Assistant: Modern Paris is a global center for...
```

--------------------------------

### API Response: Compaction Summary

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Example response after triggering manual compaction, detailing messages compacted and token savings.

```text
Compacted 40 messages into summary. Messages: 80 -> 41, Tokens: 8234 -> 4521 (saved 3713 tokens)
```

--------------------------------

### Example: Chat History After Compaction

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Displays the chat history after compaction, showing a summarized initial part and recent messages.

```text
1. Assistant: [Summary of previous conversation]: Discussion covered French capital Paris,
   its history from Roman times, major landmarks, cultural significance...
41. User: What about modern Paris?
42. Assistant: Modern Paris is a global center for...
```

--------------------------------

### /compact/clear (GET)

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Clears the entire conversation history, effectively starting a new chat session.

```APIDOC
## GET /compact/clear

### Description
Clear the conversation history and start fresh. This resets the chat memory to an empty state.

### Method
GET

### Endpoint
`/compact/clear`

### Request Example
```bash
curl http://localhost:8080/compact/clear
```
```

--------------------------------

### Run Spring Boot Application

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/CLAUDE.md

Runs the Spring Boot application using the Maven wrapper. This command starts the embedded server and makes the application accessible.

```bash
./mvnw spring-boot:run

```

--------------------------------

### /compact/memory (GET)

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Enables chat with the compacting memory feature. Sends a user message and receives a response, with older messages being compacted to save tokens.

```APIDOC
## GET /compact/memory

### Description
Chat with compacting memory enabled. This endpoint processes user messages, summarizing older parts of the conversation to stay within token limits while retaining essential context.

### Method
GET

### Endpoint
`/compact/memory`

### Parameters
#### Query Parameters
- **message** (string) - Required - Your message to the AI.

### Request Example
```bash
curl "http://localhost:8080/compact/memory?message=Tell+me+about+Spring+AI"
```
```

--------------------------------

### /compact/trigger (GET)

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Manually triggers the compaction of the conversation history. Returns a summary of the compaction process, including message and token savings.

```APIDOC
## GET /compact/trigger

### Description
Manually trigger compaction of conversation history. This operation summarizes the oldest messages in the chat memory to reduce token count.

### Method
GET

### Endpoint
`/compact/trigger`

### Response
#### Success Response (200)
- **Summary** (string) - A message detailing the number of messages compacted and token savings achieved.

### Response Example
```
Compacted 40 messages into summary. Messages: 80 -> 41, Tokens: 8234 -> 4521 (saved 3713 tokens)
```

### Request Example
```bash
curl http://localhost:8080/compact/trigger
```
```

--------------------------------

### /memory (GET)

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Standard chat endpoint using Spring AI's default `MessageChatMemoryAdvisor` for comparison purposes. This endpoint does not use message compaction.

```APIDOC
## GET /memory

### Description
Standard chat endpoint using Spring AI's default `MessageChatMemoryAdvisor`. This is provided for comparison and does not utilize the message compaction feature.

### Method
GET

### Endpoint
`/memory`

### Parameters
#### Query Parameters
- **message** (string) - Required - Your message to the AI.

### Request Example
```bash
curl "http://localhost:8080/memory?message=Hello"
```
```

--------------------------------

### Add Google GenAI Dependency

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Add the Google GenAI starter dependency to your pom.xml to use a different LLM for summarization.

```xml
<dependency>
    <groupId>org.springframework.ai</groupId>
    <artifactId>spring-ai-starter-model-google-genai</artifactId>
</dependency>
```

--------------------------------

### Run All Tests

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Command to execute all tests in the project using Maven.

```bash
./mvnw test
```

--------------------------------

### Build Application with Maven

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/CLAUDE.md

Builds the Spring Boot application using Maven. Ensures all dependencies are downloaded and the project is compiled.

```bash
./mvnw clean install

```

--------------------------------

### Run All Tests with Maven

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/CLAUDE.md

Executes all tests defined in the project using Maven. This is useful for verifying the application's correctness.

```bash
./mvnw test

```

--------------------------------

### Configure Compacting Memory Properties

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Configure the behavior of the CompactingChatMemoryAdvisor in application.properties. Adjust these values based on your needs.

```properties
compact.memory.max-messages=100
compact.memory.compact-threshold=80
compact.memory.messages-to-compact=40
```

--------------------------------

### API Call: Chat with Compacting Memory

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Initiate a chat session with the compacting memory enabled by sending a message to this endpoint.

```bash
curl "http://localhost:8080/compact/memory?message=Tell+me+about+Spring+AI"
```

--------------------------------

### Compacting Chat Memory Architecture

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Illustrates the flow of a user message through the CompactingChatMemoryAdvisor, including the compaction process.

```text
User Message → CompactingChatMemoryAdvisor
                ├─ Check if compaction needed (message count ≥ threshold?)
                ├─ If yes: Summarize oldest messages via LLM
                ├─ Add message to chat memory
                ├─ Retrieve full conversation history
                └─ Pass to ChatModel for response
```

--------------------------------

### Configure Multiple LLMs

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Configure both the primary chat model and the summarization model in application.properties. This allows for cost and performance optimization.

```properties
# Primary chat model (OpenAI GPT-5)
spring.ai.openai.api-key=${OPENAI_API_KEY}
spring.ai.openai.chat.options.model=gpt-5
spring.ai.openai.chat.options.temperature=1.0

# Summarization model (Gemini 2.5 Flash - cheaper/faster)
spring.ai.google.genai.api-key=${GOOGLE_GENAI_API_KEY}
spring.ai.google.genai.chat.options.model=gemini-2.5-flash
```

--------------------------------

### Run Single Test

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Command to run a specific test class using Maven, identified by its name.

```bash
./mvnw test -Dtest=CompactingChatMemoryAdvisorTest
```

--------------------------------

### Run a Specific Test with Maven

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/CLAUDE.md

Runs a single test class or method using Maven. This is helpful for focused testing during development.

```bash
./mvnw test -Dtest=ClassName#methodName

```

--------------------------------

### Set OpenAI API Key

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Set the OPENAI_API_KEY environment variable before running the application. This is required for OpenAI integration.

```bash
export OPENAI_API_KEY=your-api-key-here
```

--------------------------------

### Manually Trigger Compaction

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Manually trigger the compaction process for the chat memory. This forces summarization of older messages.

```bash
curl http://localhost:8080/compact/trigger
```

--------------------------------

### Send Message to Compacting Memory

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Send a message to the endpoint using the CompactingChatMemoryAdvisor, which summarizes old messages to preserve context.

```bash
curl "http://localhost:8080/compact/memory?message=Hello"
```

--------------------------------

### Create Separate ChatModel Bean for Summarization

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Define a separate ChatModel bean, qualified for the summarization LLM (e.g., Gemini), in your Spring configuration. This bean will be injected into the CompactingChatMemoryAdvisor.

```java
@Bean
@Qualifier("geminiChatModel")
public ChatModel geminiChatModel(GoogleGenAiChatModel googleGenAiChatModel) {
    return googleGenAiChatModel;
}

@Bean
public CompactingChatMemoryAdvisor compactingChatMemoryAdvisor(
        ChatMemory compactingChatMemory,
        @Qualifier("geminiChatModel") ChatModel geminiChatModel,  // Use Gemini for summaries
        CompactingMemoryProperties properties) {
    return new CompactingChatMemoryAdvisor(
            compactingChatMemory,
            geminiChatModel,  // Summaries use cheap/fast Gemini Flash
            properties.maxMessages(),
            properties.compactThreshold(),
            properties.messagesToCompact()
    );
}
```

--------------------------------

### Send Message to Standard Memory

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Send a message to the endpoint using standard memory, which discards old messages. Used for comparison.

```bash
curl "http://localhost:8080/memory?message=Hello"
```

--------------------------------

### Clear Conversation History

Source: https://github.com/danvega/compacting-chat-memory-advisor/blob/master/README.md

Clear all conversation history from the memory. Use this to reset the chat context.

```bash
curl http://localhost:8080/compact/clear
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.