Files
Pamela Chia 770f1c2b06 fix(aeo): remove ua-based markdown serving (#47770)
## Summary
The `ChatGPT-User` live-fetch agent's user-facing reader hard-fails
(`(400) OK`) on pages we serve it as markdown via user-agent matching,
which made supabase.com blog and product pages unreadable in that
assistant. I root-caused this with a controlled fetch diagnostic
cross-checked against our request logs: the failing fetches never reach
our origin (the failure is cached on their side), pages served as plain
HTML read fine everywhere we tested, and the same failure reproduces on
other major sites that serve UA-matched markdown, so the reader bug is
upstream.

This PR removes user-agent-based markdown serving entirely rather than
special-casing one agent: UA sniffing is a guess about contractless
clients whose fetchers change without notice, and this incident showed
the failure mode is silent (we keep serving 200s while the user-facing
agent breaks). Markdown remains available on every explicit signal —
`Accept: text/markdown` q-value negotiation, explicit `.md` URLs, and
llms.txt — which is the same contract-driven model the Claude fetcher
already uses successfully (it sends `Accept: text/markdown, text/html,
*/*` and keeps receiving markdown after this change).

## Changes
- Remove the `LLM_USER_AGENT` regex and the `userAgent` parameter from
`negotiateMarkdown` in `packages/common/markdown-negotiation.ts`;
decisions now depend only on `Accept`, the `.md` suffix, and the
markdown-variant manifest
- Update both consuming middlewares (`apps/www`, `apps/docs`) to the new
signature; no behavior change for Accept-negotiated or `.md` requests
- Add the missing `Vary: Accept` header to docs guides-md 200 responses
(the www `api-v2/md` route already declares it)
- Fix a pre-existing www bug surfaced in review: explicit changelog
`.md` URLs rewrote to a doubled `.md.md` path (404) under a
markdown-preferring `Accept`, and 406'd on a non-matching `Accept`. The
www middleware now strips the `.md` suffix before slug lookup and passes
`isMarkdownSuffix` into `negotiateMarkdown`, folding the separate
`MD_PAGES` `.md` block into the single negotiation path (same shape as
the docs middleware)
- Rework tests: UA-independence suites replace the per-agent rewrite
tests; a probe Accept header now 406s regardless of user agent
(previously agent UAs were exempt); new changelog `.md` negotiation
coverage

## Testing
Tested locally:
- [x] www middleware suite 36/36, docs middleware suite 17/17
- [x] typecheck green for common, www, docs

Verified on the Vercel previews (www + docs) with curl:
- [x] `ChatGPT-User` and `Claude-User` UA GETs on blog/pricing/guide
pages return `text/html` with a default Accept
- [x] Claude's real Accept (`text/markdown, text/html, */*`) still
returns `text/markdown`; `Accept: text/markdown` and `.md` URLs return
`text/markdown`; probe Accept returns 406
- [x] `/changelog/<slug>.md` with `Accept: text/markdown` returns the
entry markdown as a direct 200 (production today detours through a 308
to the bare URL); changelog index `.md` and bare-entry Accept
negotiation also verified
- [x] docs guides markdown 200s carry `Vary: Accept`

The intermediate commit (ChatGPT-User-only exclusion) was already
verified on the preview: `ChatGPT-User` got HTML while
`Accept`/`.md`/other-UA markdown was unaffected.

Expected effects post-merge: UA-driven markdown volume in the request
logs (~92% of md traffic) collapses to the Accept + `.md` baseline;
named-agent page requests return to prerendered/static serving,
reversing the extra Vercel function invocations the UA rewrite
introduced; user-facing readability in the affected assistant recovers
within ~24h as its fetch cache revalidates. The md-share dashboard gets
a dated annotation; the ratio is not comparable across this change.

## Linear
- fixes GROWTH-973


<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->

## Summary by CodeRabbit

* **Bug Fixes**
* Markdown and HTML routing now depends on the request’s `Accept` header
and `.md` links, making content negotiation more predictable.
* Requests that don’t accept available content now consistently return
`406 Not Acceptable`, even for bot-like user agents.
* Guide markdown responses now include an `Accept`-based cache variation
header to improve correct caching behavior.

<!-- end of auto-generated comment: release notes by coderabbit.ai -->
2026-07-10 13:50:34 +08:00
..
2026-07-01 12:59:00 +02:00
2026-07-03 15:00:43 +10:00
2026-02-23 15:47:03 +00:00

supabase.com

Overview

Refer to the Development Guide to learn how to run this site locally.

To get started copy the example env file using cp .env.local.example .env.local.

Best practices

Images

  • Resize images: All new images should be resized to only the maximum resolution needed when rendering on the frontend. Don't upload images larger than what will be displayed.
  • Compress images: All new images should be compressed before committing. Use tools like Clop or ImageOptim to reduce file size without noticeable quality loss.
  • Image locations: Store blog post images in apps/www/public/images/blog/. Event images go in apps/www/public/images/events/.

OG image generation

Open Graph (OG) images for social sharing are handled differently across content types:

  • Blog posts: Use static images via imgSocial and imgThumb fields (see Blog posts section below)
  • Events: Use dynamic generation via Edge Function (with optional og_image override)
  • Customer stories: Use dynamic generation via Edge Function (no static option)

The og-images Edge Function (supabase/functions/og-images/) automatically generates OG images for events and customer stories. It's deployed via .github/workflows/og_images.yml when changes are made to the function code.

Development: In local development, the function runs at http://127.0.0.1:54321/functions/v1/og-images. Ensure Supabase is running locally (supabase start).

Content frontmatter image fields

Different content types use different image field conventions:

Blog posts

Blog posts support two image fields in their frontmatter:

  • imgSocial: Used for Open Graph and social media sharing (X, LinkedIn, etc.). These images should include text overlays since they appear standalone in social feeds without accompanying text.
  • imgThumb: Used for internal thumbnails displayed on the blog listing pages and featured posts. These images don't need text overlays since they're always displayed alongside the post title and description.

This naming convention was introduced to replace the previously confusing thumb and og fields, which were often mixed up. The new names clearly indicate:

  • imgSocial: Purpose-built for social media sharing (needs text overlays)
  • imgThumb: Optimized for site display (clean, no text overlays)

This separation allows you to optimize images for their specific use case while maintaining a clear, unambiguous naming convention.

Image path format

Always use relative paths (just the filename or subfolder/filename). The /images/blog/ prefix is added automatically by the code.

  • ✅ Correct: imgSocial: my-post/og.png or imgSocial: og.png
  • ❌ Wrong: imgSocial: /images/blog/my-post/og.png (creates double prefix)

Image fallback behavior

For site display (what visitors see):

  • Priority 1: imgThumb
  • Priority 2: imgSocial (if imgThumb is missing)
  • Priority 3: /images/blog/blog-placeholder.png (if both missing)

For social sharing (Open Graph meta tags):

  • Priority 1: imgSocial
  • Priority 2: imgThumb (if imgSocial is missing)
  • Priority 3: No fallback (undefined)

What happens if fields are not provided?

  • If only imgThumb is provided: Site displays the image correctly, social sharing uses imgThumb as fallback
  • If only imgSocial is provided: Social sharing uses it, site display uses it as fallback
  • If neither is provided: Site shows placeholder image, social sharing has no image
  • Best practice: Provide both fields for optimal display and social sharing

Example

---
title: 'My Blog Post'
imgSocial: 2025-01-01-my-post/og.png # Relative path - with text overlay for social sharing
imgThumb: 2025-01-01-my-post/thumb.png # Relative path - without text, clean image
---

The images would be stored at:

  • apps/www/public/images/blog/2025-01-01-my-post/og.png
  • apps/www/public/images/blog/2025-01-01-my-post/thumb.png

Or if using the same image for both:

---
title: 'My Blog Post'
imgSocial: my-image.png # Stored at: apps/www/public/images/blog/my-image.png
imgThumb: my-image.png
---

Events

Events use different image fields to avoid confusion with their display patterns:

  • thumb: Used for event grid item thumbnails (small cards in listing)
  • cover_url: Used for the featured event banner (large display on events page)
  • og_image (optional): Used to override the dynamically generated OG image for social sharing

OG image generation

Events automatically generate Open Graph images using the og-images Supabase Edge Function. The function creates images dynamically based on:

  • Event type (conference, hackathon, etc.)
  • Title (or meta_title if provided)
  • Description (or meta_description if provided)
  • Date (formatted as "DD MMM YYYY" using the event's timezone)
  • Duration (if provided)

If you need a custom OG image that differs from the auto-generated one, you can provide an og_image field in the frontmatter. This will override the dynamic generation.

Example:

---
title: 'Supabase Meetup'
thumb: /images/events/2025-01-meetup/thumbnail.png
cover_url: https://external-cdn.com/event-banner.jpg
og_image: /images/events/2025-01-meetup/custom-og.png # Optional override
---

Note: The og_image field is optional. If not provided, OG images are generated automatically via the Edge Function.

Go pages (/go/*)

/go/ is a system for building standalone campaign landing pages (lead generation, legal, thank-you flows). The name is intentionally generic — these pages are typically linked from ads, emails, or partner campaigns and are not part of the main site navigation.

Pages are defined as TypeScript objects (not MDX files) and validated against Zod schemas at build time.

Parts

Location Purpose
apps/www/_go/ Page definitions. Each file exports a page object. index.tsx registers all pages.
apps/www/app/go/[slug]/page.tsx App Router route — renders the page for a given slug, handles 404s and metadata.
apps/www/components/Go/GoPageRenderer.tsx www-specific wrapper — adds the Supabase logo header and footer, registers custom section renderers.
packages/marketing/src/go/ Framework-agnostic core: schemas, section components, templates, form server action.
packages/marketing/src/crm/ CRM client abstraction (HubSpot + Customer.io) used by the form server action.

Page structure

Each page specifies a template which determines its top-level layout:

  • lead-gen — hero + arbitrary sections (form, metrics, feature grid, tweets, social proof, etc.)
  • thank-you — hero + sections + confetti animation
  • legal — hero + table-of-contents sidebar + markdown body

Pages are arrays of typed section objects. The SectionRenderer in packages/marketing dispatches each section to the right component based on its type field.

Custom renderers

The marketing package doesn't know about topTweets data or the Pages Router basePath, so the tweets section type has no default renderer. GoPageRenderer.tsx in www registers TweetsSection as a custom renderer for that type. This is the extension point for any section that requires www-specific dependencies.

Adding a new page

  1. Create a new file in apps/www/_go/<category>/my-page.tsx exporting a page object.
  2. Register it in apps/www/_go/index.tsx.
  3. The page will be available at /go/<slug> automatically via static generation.

Customer Stories (Case Studies)

Customer stories are defined in MDX files (apps/www/_customers/*.mdx) and use a different approach:

  • No og_image field: Customer stories do NOT use static OG images
  • Dynamic OG generation: All customer story OG images are automatically generated using the og-images Supabase Edge Function
  • The function creates images based on the customer slug and title (or meta_title if provided)

Do not include an og_image field in customer story frontmatter. It will be ignored. OG images are always generated dynamically.

Example (in apps/www/_customers/company-abc.mdx):

---
name: Company ABC
title: Company ABC built their platform with Supabase
# DO NOT include og_image - it's generated automatically
logo: /images/customers/logos/company-abc.png
---

Legacy Case Studies (in data/CustomerStories.ts):

  • imgUrl: Path to the case study image in the source data
  • This gets mapped to imgThumb when rendered via BlogGridItem component
  • Case studies only need one image for site display (no separate social sharing image)

Example (in data/CustomerStories.ts):

{
  type: 'Customer Story',
  title: 'Company ABC built their platform with Supabase',
  description: '...',
  organization: 'Company ABC',
  imgUrl: 'images/customers/logos/company-abc.png', // Full path from public/
  logo: '/images/customers/logos/company-abc.png',
  url: '/customers/company-abc',
}