How git-eat/giteat works
Project type and purpose
• This is a full‑stack, microservice‑style web application.
• Purpose: An AI‑powered code‑review service (“GIT‑EAT”) integrated with GitLab, providing automated code feedback, notifications, dashboards, performance metrics, and secured via OAuth/JWT.
File‑structure analysis
• Top‑level separation into:
• frontend (React + Vite SPA)
• backend (five independent Spring Boot modules: ai, notification, report, rest, security)
• exec (deployment/CI‑CD docs, DB dumps)
• Each backend module contains its own build (Gradle), Dockerfile, application.properties, controllers/services/entities/repositories.
• The “rest” module exposes the main user‑facing API endpoints (PRs, repos, comments, stats, webhooks).
• The “security” module handles OAuth interception, JWT issuance, and secures all service calls.
• “ai” module runs an asynchronous daemon to generate AI reviews via GitLab and an LLM.
• “notification” module sends messages to external Mattermost.
• “report” module fetches and stores Lighthouse performance data.
• CI/CD artifacts: Dockerfiles in each module, Jenkins and NGINX configurations in exec/docs.
README insights
• OAuth login via GitLab.
• Webhook integration to sync MR events.
• Automatic AI review creation upon MR events.
• GitLab API integration for posting review comments.
• Data persistence in MySQL.
• Frontend dashboard, MR detail pages, performance measurement (Lighthouse).