Hi PH! I'm building AimCode — an AI editor that verifies its own code

Solo dev here. Two things I've been building lately:

1. AimCode — a VS Code fork where the AI writes code and a built-in SDK verifies it.

Auto-checks every turn, an independent audit before anything ships (kill port, wipe

data, restart baseline), and FAIL means the settlement is downgraded to HOLD. Latest

experiment: one English spec → a playable full-stack game in 23 minutes, 73 files,

zero bugs on first run.

2. A Microsoft Store IAP bridge for Electron apps. Turns out consumable IAP for

non-C#/UWP apps is completely undocumented — no public end-to-end path exists. We

shipped one in production anyway: native N-API host embedding .NET 10, a WinForms

message pump so purchase dialogs actually show, server-side validation through the

collections API, and an idempotent transaction ledger. Learned the hard way that

"non-Active entitlement" ≠ refund for consumables.

The question behind both: how do we trust what AI builds, and how do we sell what

AI ships? Currently my answer is tooling, not vibes.

What's your workflow for verifying AI-written code? Genuinely curious how others

handle this.