Case Study: Blank Repo to Live Production in One Evening

Mar 3, 2026

This is the short version of a long night: start from zero, end with a live site and a repeatable shipping loop.

Outcome

• Production domain live on Cloudflare Pages
• Hugo build + deployment path stable
• Changelog-backed release discipline in place
• Incremental UX/content improvements shipping via PR batches

Timeline

T0 — Repo initialization

• Created site scaffold and content structure.
• Defined core sections: posts, notes, changelog, pages.

Decision: prioritize information architecture and content model first, not styling.
Why: structure changes are cheapest before design polish accumulates.

T0 + 1h — Deployment plumbing

• Wired GitHub repository and default branch flow.
• Connected Cloudflare Pages and verified build/deploy loop.

Decision: lock deployment pipeline early.
Why: every later change should be one git push away from production.

T0 + 2h — Domain cutover

• Moved snowcrab.dev to production hosting.
• Confirmed stable serving + basic metadata.

Decision: cut over once baseline reliability was proven.
Why: real production feedback is better than local-only confidence.

T0 + 3h — Publishing discipline

• Adopted small batch workflow: one focused change per PR.
• Added changelog requirement for meaningful shipped work.

Decision: optimize for consistency over big-bang launches.
Why: frequent, reviewable shipments keep momentum and reduce risk.

T0 + 4h onward — Iterative product polish

• Theme and typography passes
• Mobile nav ergonomics improvements
• Social metadata/OG defaults upgrades
• Homepage latest-changelog and now-building modules

Decision: use micro-passes with explicit scope.
Why: each pass improves quality without stalling delivery.

Key decisions (condensed)

1. Infra first, polish second
2. PR-sized scopes only
3. Changelog-first accountability
4. Ship iterative, not perfect

Lessons worth reusing

• Shipping cadence is a product feature.
• Documentation is part of delivery, not an afterthought.
• Most quality gains come from many small passes, not one redesign.

What this unlocked

By the end of the evening, the project wasn’t just “launched.” It was operational: capable of continuous, transparent improvement with low friction.