The cost figures only cover the runs that reached a working state. They leave out every later pass that still has to reload the full graph once a dependency or schema shift forces another round of cross file fixes. The installer comparison stops at the first output. It does not show what happens when the generated code later collides with an upstream change that the starter itself does not absorb
Valid points on both fronts — especially on the hidden cost of context reloading during refactors.
On AI costs: You're 100% right. $200 only covers reaching the initial green build. The moment you hit a schema migration or cross-file refactor, feeding the whole dependency graph back into the context window causes token burn to compound rapidly.
On upstream shifts & maintenance: This is a real issue with monolithic, tightly-coupled starters. I designed Codapult around an adapter-first architecture partly for this reason — as well as making provider swapping seamless. Third-party integrations (Auth, Payments, DB) are isolated behind strict TypeScript interfaces.
If a provider introduces a breaking API change, you only refactor a single adapter implementation (e.g., stripe-adapter.ts). Your core application logic and UI remain completely untouched, keeping the blast radius contained to a single module instead of cascading through the entire repo.
Andrew
The cost figures only cover the runs that reached a working state. They leave out every later pass that still has to reload the full graph once a dependency or schema shift forces another round of cross file fixes. The installer comparison stops at the first output. It does not show what happens when the generated code later collides with an upstream change that the starter itself does not absorb