Fix AI prompt wiring, asset generation flow, and signup trigger

[jadru]

Summary

• wire headline and OG prompt generation through the new structured prompt config pipeline
• fix the wizard asset-generation flow so preview advances correctly and duplicate generation requests are blocked client/server side
• generate and layer current Supabase types, and fix the auth signup trigger to populate public.users from auth.users.email

Verification

• yarn tsc --noEmit
• yarn vitest run tests/lib/ai/prompt-config.test.ts
• yarn build
• applied Supabase migrations through 013_fix_handle_new_user_email.sql and verified remote migration alignment
• browser smoke: headline generation + asset generation completed with exactly one /api/assets/generate request
• signup trigger verification: public signup and admin create both create matching public.users rows; auth.users / public.users mismatch query returned zero rows

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
brand-kit	Ready	Preview, Comment	Mar 10, 2026 9:03am

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 10c36de9f5

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Allow regeneration when status is already completed

startGeneration now exits immediately when the first status poll returns completed, which prevents the subsequent /api/assets/generate call from ever being sent. In the current UI, the preview screen always exposes a retry/regenerate action, so retrying a previously completed project will no longer trigger a new generation and will silently keep the old assets.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Move generating status update after prerequisite checks

The project is marked status='generating' before validating dependent records (for example, the style preset lookup just below). If that lookup fails (missing/deleted preset or query error), the handler throws before entering the pipeline try/catch that resets status, leaving the row stuck as generating; later generate attempts then return ASSET_GENERATION_IN_PROGRESS even though no pipeline is running.

Useful? React with 👍 / 👎.