Developing a Spring AI Enhanced Restaurant Booking System Employing an API-first Approach

This multi-module project hosts a client code-generated from an OpenAPI derivative of the ResOs API combined with a Spring AI implementation. It also includes an MCP server, MCP client configuration for use with Claude and a standalone ReactJS powered chatbot UI.

• Background
• Getting started
• Prerequisites
• How to
  • Clone
  • Build
  • Consume
  • Run
• Documentation
  • Architecture - Comprehensive architecture documentation
  • Planning - Roadmap and future work
  • Project Origin - How this project started
• Also see
  • ResOS API - External API reference

Background

As a Spring Boot and Spring AI developer, I want to consume libraries that make it convenient to add capabilities to my application(s) as for the following

Use-case:

• Imagine instead of using OpenTable or Tock you could converse with a chatbot to search for restaurant(s) and make reservation(s) on your behalf.

Technologies

• Spring Boot 4.0.1
• Spring AI 2.0.0-M1
• Spring Cloud 2025.1.0
• Spring Security 7.0.2
• Java 25
• Maven 3.9.11

Getting started

Start with:

• A Github account
• (Optional) An API key from ResOS
  • you only need one if you intend to register as a restaurateur!
  • we will spin up a backend that is API-compatible, implemented with Spring Boot Starter Data JDBC
• An LLM provider
  • e.g., Groq Cloud, OpenRouter, or OpenAI

Prerequisites

• Git CLI (2.43.0 or better)
• Github CLI (2.65.0 or better)
• httpie CLI (3.2.2 or better)
• Java SDK (25 or better)
• Maven (3.9.11 or better)
• an LLM provider account (if using public cloud or commercially hosted models)

How to clone

with Git CLI

git clone https://github.com/pacphi/spring-ai-resos

with Github CLI

gh repo clone pacphi/spring-ai-resos

How to build

Open a terminal shell, then execute:

cd spring-ai-resos
mvn clean install

How to consume

If you want to incorporate any of the starters as dependencies in your own projects, you would:

Add dependency

Maven

<dependency>
    <groupId>me.pacphi</groupId>
    <artifactId>spring-ai-resos-client</artifactId>
    <version>{release-version}</version>
</dependency>

Gradle

implementation 'me.pacphi:spring-ai-resos-client:{release-version}'

Replace occurrences of {release-version} above with a valid artifact release version number

Add configuration

Following Spring Boot conventions, you would add a stanza like this to your:

application.properties

default.url=${RESOS_API_ENDPOINT:https://api.resos.com/v1}

application.yml

default:
  url: ${RESOS_API_ENDPOINT:https://api.resos.com/v1}

To activate the client, specify an API key (if required), and tune other associated configuration.

Consult the chatbot module's configuration for alternative dependencies and configuration that are available to add.

Configuration will be found in labeled spring.config.activate.on-profile sections of the application.yml file.

How to run

You're going to need to launch the backend module first, unless you're a restaurateur, and you have a valid API key for interacting with the ResOS v1.2 API.

To launch the backend, open a terminal shell and execute

cd backend
mvn clean spring-boot:run -Dspring-boot.run.profiles=dev -Dspring-boot.run.jvmArguments="--add-opens java.base/java.net=ALL-UNNAMED"

There's the chatbot module.

But there's also a way to integrate with Claude desktop via MCP client configuration that will consume the MCP server implementation.

with Claude Desktop

Claude Desktop can connect to the MCP server using STDIO transport. This allows Claude to directly invoke restaurant management tools.

Prerequisites

1. Get the STDIO variant of the MCP server:
   • Option A: Download from Releases - Download spring-ai-resos-mcp-server-stdio-{VERSION}.jar from the Releases page.
   • Option B: Build from Source - Run cd mcp-server && mvn clean package -Pstdio
2. Ensure the backend is running (if using local development):

cd backend
mvn spring-boot:run -Dspring-boot.run.profiles=dev -Dspring-boot.run.jvmArguments="--add-opens java.base/java.net=ALL-UNNAMED"

Configuration

Add the following to your Claude Desktop configuration file:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json
• Linux: ~/.config/claude/claude_desktop_config.json

If using downloaded release JAR:

{
  "mcpServers": {
    "spring-ai-resos": {
      "command": "java",
      "args": [
        "-Dspring.profiles.active=stdio",
        "-jar",
        "<path-to-jar>/spring-ai-resos-mcp-server-stdio-{VERSION}.jar"
      ],
      "env": {
        "RESOS_API_ENDPOINT": "http://localhost:8080/api/v1/resos"
      }
    }
  }
}

If built from source:

{
  "mcpServers": {
    "spring-ai-resos": {
      "command": "java",
      "args": [
        "-Dspring.profiles.active=stdio",
        "-jar",
        "<path-to-project>/mcp-server/target/spring-ai-resos-mcp-server-1.0.0-SNAPSHOT.jar"
      ],
      "env": {
        "RESOS_API_ENDPOINT": "http://localhost:8080/api/v1/resos"
      }
    }
  }
}

Replace <path-to-jar> or <path-to-project> with the absolute path to your JAR file or project directory.

Available Tools

Once connected, Claude Desktop will have access to these tools:

Tool	Description
getTables	Fetch all restaurant tables
getCustomers	Fetch customer records with filtering/pagination
getCustomerById	Fetch a specific customer
getFeedback	Fetch customer feedback and reviews
getFeedbackById	Fetch specific feedback
getOpeningHours	Fetch opening hours for the next two weeks
getOpeningHoursById	Fetch specific opening hours

Verification

1. Restart Claude Desktop after updating the configuration
2. Look for the tools icon (hammer) in the Claude interface
3. You should see "spring-ai-resos" listed with available tools
4. Try asking: "Show me all customers" or "What tables are available?"

Troubleshooting

Server not connecting:

• Verify the JAR path is absolute and correct
• Ensure Java 25+ is installed and in PATH
• Check that the backend server is running on port 8080

Tools not appearing:

• Verify the configuration JSON syntax is valid
• Check Claude Desktop logs for errors
• Restart Claude Desktop completely (not just the chat)

Backend connection errors:

• Ensure RESOS_API_ENDPOINT environment variable is correct
• Verify the backend is accessible at the configured URL

with Chatbot

Follow these instructions.

To launch the server module, open a terminal shell and execute

cd mcp-server
export RESOS_API_ENDPOINT=http://localhost:8080/api/v1/resos
mvn spring-boot:run -Dspring-boot.run.profiles=cloud,dev

Next, we'll store an API key in a credential file that will allow the chatbot to interact with an LLM service provider.

cd ../mcp-client

leveraging OpenAI

Build and run a version of the chatbot that is compatible for use with OpenAI. You will need to obtain an API key.

Before launching the app:

• Create a config folder which would be a sibling of the src folder. Create a file named creds.yml inside that folder. Add your own API key into that file.

spring:
  ai:
    openai:
      api-key: { REDACTED }

Replace {REDACTED} above with your OpenAI API key

Next, to launch the chatbot, open a terminal shell and execute

mvn spring-boot:run -Dspring-boot.run.profiles=openai,dev

leveraging Groq Cloud

Build and run a version of the chatbot that is compatible for use with Groq Cloud. You will need to obtain an API key. Note that Groq does not currently have support for text embedding. So if you intend to run with the groq-cloud Spring profile activated, you will also need to provide additional credentials

Before launching the app:

• Create a config folder which would be a sibling of the src folder. Create a file named creds.yml inside that folder. Add your own API key into that file.

spring:
  ai:
    openai:
      api-key: { REDACTED-1 }
      embedding:
        api-key: { REDACTED-2 }

Replace {REDACTED-1} and {REDACTED-2} above with your Groq Cloud API and OpenAI keys respectively.

Next, to launch the chatbot, open a terminal shell and execute

mvn spring-boot:run -Dspring-boot.run.profiles=groq-cloud,dev

leveraging OpenRouter

Build and run a version of the chatbot that is compatible for use with OpenRouter. You will need to obtain an API key. Note that OpenRouter does not currently have support for text embedding. So if you intend to run with the openrouter Spring profile activated, you will also need to provide additional credentials

Before launching the app:

• Create a config folder which would be a sibling of the src folder. Create a file named creds.yml inside that folder. Add your own API key into that file.

spring:
  ai:
    openai:
      api-key: { REDACTED-1 }
      embedding:
        api-key: { REDACTED-2 }

Replace {REDACTED-1} and {REDACTED-2} above with your OpenRouter API and OpenAI keys respectively.

Next, to launch the chatbot, open a terminal shell and execute

mvn spring-boot:run -Dspring-boot.run.profiles=openrouter,dev

Now, visit http://localhost:8081 in your favorite web-browser.