springai-mcp-server1

Minimal Spring AI MCP Server sample.

Testing (Windows - cmd primary, PowerShell secondary)

Prerequisites:

• Java 17 installed
• Use the included Gradle wrapper (gradlew.bat) from project root

1. Build the application (cmd):

gradlew.bat bootJar

PowerShell equivalent:

./gradlew.bat bootJar

2. Run tests (cmd):

gradlew.bat test

PowerShell equivalent:

./gradlew.bat test

3. Start the application (cmd, background):

gradlew.bat bootRun

PowerShell equivalent (foreground):

./gradlew.bat bootRun

4. Send a test request (cmd - primary):

curl -X POST "http://localhost:8080/mcp/message" -H "Content-Type: application/json" -d "{\"id\":\"1\",\"type\":\"policy.check\",\"payload\":{}}"

PowerShell alternative:

$body = '{"id":"1","type":"policy.check","payload":{}}'
Invoke-RestMethod -Method Post -Uri http://localhost:8080/mcp/message -Body $body -ContentType 'application/json'

Expected response example:

{"id":"1","status":"ok","result":{"decision":"approve","details":"mocked"}}

Tips:

• If the server isn't ready when you POST, wait a few seconds and retry.
• To build a Docker image:

gradlew.bat bootJar
docker build -t springai-mcp:local .
docker run -p 8080:8080 springai-mcp:local

Contact: See project files for implementation details.

Manual testing (no commands)

• Build artifacts: confirm build/libs/ contains a generated JAR after a successful build.
• Application startup: verify logs show Spring Boot banner and a line like "Started Application" and "Tomcat" or "Started" indicating the server is listening on port 8080.
• Health check: the /actuator/health endpoint (if enabled) or the root logs should indicate the app is up.
• Endpoint availability: confirm /mcp/message is exposed by inspecting application startup logs that show registered mappings (e.g., "Mapped "POST /mcp/message"").
• Request/response flow: when sending a POST to /mcp/message, expect a JSON response with status: ok and a result.decision field; verify the value matches the adapter (mock returns approve).
• Persistence: check H2 console or application logs for SQL insert statements showing rows inserted into the decision_entity table (or inspect build/test output for repository activity).
• Error handling: provide a malformed request body and confirm the app returns an HTTP 4xx/5xx with an explanatory message in logs.

Notes:

• These manual checks require viewing application logs, generated files, or the database console rather than running scripted commands.
• If you want, I can add log examples and a short troubleshooting checklist.

GUI step-by-step (Windows) — double-click / menu actions

Option A — Using IntelliJ IDEA (recommended, no typing required):

1. Open IntelliJ IDEA.

2. Click File → Open... and double-click the project folder springai-mcp-server1 to open it.

3. Wait for the IDE to import the Gradle project (watch the status bar).

4. Open the Gradle tool window (right side). Expand Tasks → build and double-click bootJar to run the build task; watch the Run tool window for progress and logs.

5. When the build completes, open File Explorer and navigate to build\libs\ and double-click the generated .jar file (e.g. springai-mcp-server1-0.0.1-SNAPSHOT.jar) to start the app. If double-clicking opens a console that closes immediately, instead in IntelliJ use the Run tool window's run configuration to start the application.

6. In IntelliJ's Run tool window you will see Spring Boot startup logs. Confirm a line like Started Application and that the server is listening on 8080.

7. To send a request from VS Code (instead of Postman):

   • Open VS Code.
   • Click File → Open Folder... and choose the springai-mcp-server1 folder.
   • Install these extensions if prompted: Java Extension Pack, Spring Boot Extension Pack, and REST Client

... View full README on GitHub