The duplicate robots tags are the detail that would have worried me most — a page carrying both 'index, follow' from the root layout and 'noindex' from the streamed not-found, with a 200 status, leaves the outcome entirely up to whichever tag a crawler happens to honour. The dynamicParams = false fix is elegant precisely because it moves the decision to routing, before loading.tsx can stream anything. For the database-driven case you flag as the wrong fit, would the proxy.ts check you used for the encoded-slash slugs be your preferred route there too, or would you rather call notFound() before any await on the slow path so the status is set before streaming starts?