The Great Content Restoration

It was a quiet afternoon at the rhidayat.work headquarters—or so it seemed. Beneath the surface of our shiny new educational platform, a silent crisis was brewing.

As we expanded our curriculum to include thousands of exercises from the Exercism universe, we noticed something unsettling. Some challenges were missing their “Our Solution” fields. Others had redundant instruction headers, cluttering the learning experience. The deeper we looked, the more we found.

The Discovery

We didn’t just have a few broken files. We had a systemic content regression.

Using a custom-built validator script (scripts/validate-content.ts), we performed a high-speed audit of the entire repository. The results were staggering:

• 2627 files with formatting errors.
• 2051 files missing reference solutions.
• 2575 files with duplicate instruction headers.

The manual repair of such a volume would have taken weeks of mind-numbing labor. We were staring into the abyss of content debt.

The Strategy: Automate or Perish

We decided to fight fire with fire. We didn’t just want to fix the files; we wanted to build a self-healing content pipeline.

1. The Intelligent Enricher

We upgraded our enrich-exercism.ts script. It wasn’t enough to just fetch data; the script now had to be a surgeon. It was programmed to identify missing solutions, fetch the highest-rated community alternatives from the Exercism API, and inject them seamlessly into the frontmatter.

2. The Header Surgeon

We implemented regex-based deduplication logic. Any file with redundant # Instructions headers was automatically pruned, restoring the professional, high-fidelity look our users expect.

3. The Unblocking Priority

While the massive repair was running for all 2600+ files, we identified critical blockers in the TypeScript track. Users were unable to proceed to the next challenge because the “Next” button relied on a valid solution being present. We prioritized these files, deploying a fix in minutes to keep the learning momentum alive.

The Result: 0 Errors

After a massive 30-minute processing run, our validator script finally returned the most beautiful line of code a developer can see: ✅ All files are valid!

The platform is now cleaner, more consistent, and fully unblocked. Every one of the 2600+ challenges now features:

• A verified reference solution.
• Clean, single-header instructions.
• Properly formatted community solution links.

Lessons Learned

Scaling content is as much an engineering challenge as scaling infrastructure. By treating our Markdown files as “code” that requires validation and automated repair, we’ve built a platform that isn’t just large—it’s resilient.

The great restoration is complete. Now, back to building.