AI Agents Migrate a Site to Astro

A case study of using AI coding agents to migrate Code With Andrea — a 400+ page static site built on a custom Swift generator (Publish) — to Astro, preserving 100% feature parity. Done solo over a two-week holiday with only a couple of hours a day at the keyboard, running no more than two agents in parallel inside tmux on a VPS.

The through-line is a five-stage workflow that keeps human judgment at every high-stakes decision. First, "feature parity" is turned into a migration contract — a testable description of every route, redirect, metadata field, asset, interaction, and intentional exception. A throwaway prototype proves the riskiest assumption (Astro content collections could preserve canonical URLs) before bulk migration. Content conversion is handled by a deterministic, one-way migration tool with its own test harness that fails loudly on unknown placeholder syntax rather than guessing. Verification is split into fast, focused checks for normal agent runs and a full production-parity suite (plus curated screenshots) at release gates.

The central claim: agents make implementation cheaper, but they cannot know which legacy behavior is intentional, which is obsolete, or which trade-off is acceptable to your users. The author keeps the final product and production decisions — including the cut-over — in human hands. Roughly 100 issues and 80 reviewed PRs later, the site went live (the article itself is served from the migrated Astro build).