Lectures

We will break down how the ChatGPT Store works and why “minimum necessary permissions” are simultaneously about security, UX, and the criteria for passing moderation. We’ll discuss the three layers of access (model capabilities, OAuth scopes, and MCP tools), the consequential : true flag, mapping permissions to human-readable wording for the listing, and typical mistakes (over-permissioned authorization, manifest-to-description mismatch, “hidden” actions).

In this lecture, we’ll figure out which legal pages are required to publish a ChatGPT App in the Store and why this is not a mere formality: how to choose stable public URLs, where to place the pages in Next.js (routes like /legal/privacy, /legal/terms, /support), how to connect the Privacy Policy to the real architecture (logs, retention, deletion), what must be reflected in the Terms (AI disclaimers, liability, commerce), how to organize support, and which common mistakes most often fail review.

In this lecture, we assemble a practical preflight process before submitting a ChatGPT App to the Store: we fix the safety profile (what the App does and what it never does) through a small TypeScript artifact, add a set of safety cases (in addition to golden prompts), connect requirements to systemPrompt, tool descriptions, and server‑side checks (defense‑in‑depth), and also perform a basic UI accessibility check (WCAG AA) and verify that the Privacy Policy/Terms are consistent with the actual code.

In this lecture we break down how the review cycle for publishing a ChatGPT App in the Store works: what exactly you submit for review (listing, links to Privacy Policy/ Terms, permissions, production URL), how reviewers run through scenarios, why you need test accounts and sandbox modes via a flag like REVIEW_MODE, how to respond to remarks, and how to build an internal pre-flight checklist and versioning ( semver) for iterations after release.

We’ll break down how to turn a ChatGPT App listing into part of the product: what to write in the title, subtitle, and description, how not to drown in tech jargon or overpromise, how to think through visuals and metadata localization, and how to maintain release notes and use them for promotion. At the end — a practical “draft” listing for GiftGenius, a user journey diagram, and a set of common mistakes (including mismatches between the listing and real behavior, and localization via openai/locale and _meta [ "openai/userLocation" ]).