1. High Impact: Reliability & Security
*   Adopt a Templating Engine (e.g., Jinja2): 
    *   Problem: The current build scripts (build_pages.py, prep.py) use manual string replacement (e.g., .replace('<!--TITLE-->', ...)). This is brittle and prone to errors as the site grows.
    *   Benefit: Using Jinja2 would make templates much more powerful (loops, conditionals) and, crucially, would prevent XSS vulnerabilities by automatically escaping data from your CSV files.

*   Use csv.DictReader in Python:
    *   Problem: prep.py accesses CSV columns by index (e.g., row[0]). If you add a column to dogadjaji.csv, the build will break.
    *   Benefit: Accessing by name (row['title']) makes the code resilient to data schema changes.

2. Medium Impact: Developer Experience (DX)
*   Code Linting:
    *   Benefit: Adding black or flake8 for Python and prettier for HTML/CSS would ensure consistent style across the repository.

3. Low Impact: Performance & Modernization
*   CSS Consolidation:
    *   Problem: There are many small CSS files (one per page).
    *   Benefit: While fine for a small site, consolidating these or using a preprocessor like Sass would make managing global styles easier.

*   Asset Optimization:
    *   Benefit: Implementing automated image compression (for the event posters) would reduce the final site payload.
Which of these areas would you like me to focus on first? I recommend starting with the Jinja2 migration as it solves both maintainability and security issues.