5. Working-version conclusion

The thesis addresses support for teaching DevOps practices in team projects hosted on GitLab. The problem is not missing data but its distribution across issues, branches, merge requests, discussions, commits, and CI/CD processes.

The analysis separated technically verifiable conditions from questions that require expert judgement. The result is thirteen transparent R01-R13 checks. Communication and commit-message checks use proxy signals and cannot establish work quality or unethical behaviour without manual review.

Interface verification showed that webhooks are suitable for fast change notification, while complete state requires REST API enrichment. Manual synchronisation restores initial data and gaps. The prototype does not periodically scan every project.

GitPulse presents data at course, group, project, and student level. Instructors configure active checks, thresholds, and weights. The traffic-light state prioritises attention and is not an automatic grade. Separate optional modules create a merge conflict, analyse a Dockerfile, and generate a controlled CI/CD failure.

The third submission contains a functional prototype, public instance, architecture design, and a focused technical test run. Automated tests verify prepared scenarios, not pedagogical impact. The thesis therefore does not yet claim time savings, improved student outcomes, or successful classroom use.

The next stage must compare automatic results with manual assessment on real projects and measure technical accuracy, instructor time, and feedback clarity separately. Only those data can support final contribution claims.