AI-assisted portfolio maintenance

Problem

The portfolio needed navigation corrections, stronger editorial consistency and the removal of overly specific professional content, while avoiding unrelated changes.

Inputs

• Existing public repository.
• Defined scope and editorial constraints.
• AI-assisted code inspection.
• Human review and approval.
• Git and CI validation.

Workflow

1. Select one narrowly scoped change.
2. Create a clean branch from the current main branch.
3. Ask the AI assistant to inspect before editing.
4. Restrict the files and claims that may be changed.
5. Review git status and the complete diff.
6. Commit and push the isolated change.
7. Review the pull request and automated checks.
8. Merge and validate the deployed result.

Validation

• git status --short
• git diff --check
• Review of files changed.
• Pull request CI.
• HTTP validation of affected routes.
• Manual verification after deployment.

Observed result

• Broken navigation anchors were corrected.
• Southustles navigation was aligned across EN, ES and PT.
• Specific professional case studies were removed from public deployment.
• AI Automation Workspace was rewritten with factual boundaries.

Human responsibilities

• Decide what should be public.
• Validate factual accuracy.
• Review security and confidentiality.
• Approve the final diff.
• Perform commit, push and merge operations.

Limitations

• This is not a benchmark of AI performance.
• AI output may be incorrect or incomplete.
• The workflow still requires human judgment.
• The example is specific to this static portfolio.
• No confidential workplace information is used.

Reproducible checklist

• Define one small change and its allowed files.
• Inspect the current state before editing.
• State factual, security and editorial boundaries.
• Review every changed file and the complete diff.
• Run repository checks and validate affected routes.
• Approve and publish only after human review.