Workshop 2: Sonar Agent Essentials

AI coding agents write fast. Without context, they violate your standards, ignore your architecture, and introduce security issues you won't catch until production. This second workshop in the series shows you how to guide agents from the first line and verify their output before it ever reaches a pull request.

Sonar Vortex injects your project context and constraints before the agent writes, then verifies every change in real time with SonarQube's deterministic analysis. You'll install the SonarQube Plugin for Claude Code, connect it to a SonarQube project, and watch Vortex feed repository context, architecture rules, and security standards straight into the agent's loop. Real-time verification catches bugs, vulnerabilities, and reliability issues inside that loop—so risk never compounds downstream.

From there, you'll put the SonarQube Remediation Agent to work on a real backlog. Every proposed fix runs in a sandbox, gets re-analyzed, and reaches your PR only once it passes. You see the full closed-loop workflow in action: guide agents with context, verify their output with deterministic analysis, and solve identified issues automatically.

Prerequisites:

Participants should use a personal machine with Docker, a GitHub account, and access to either Claude or Codex.

A working SonarQube instance is helpful (covered in Workshop 1) but not strictly required.