diff --git a/.agents/skills/seo/SKILL.md b/.agents/skills/seo/SKILL.md
new file mode 100644
index 0000000000..d9572a5321
--- /dev/null
+++ b/.agents/skills/seo/SKILL.md
@@ -0,0 +1,237 @@
+---
+name: seo
+description: >
+  SEO best practices for the Grida project across Next.js pages, blog posts,
+  and documentation. Covers Next.js metadata API, Open Graph / Twitter cards,
+  sitemaps, image search optimization, structured data (JSON-LD), Docusaurus
+  frontmatter, and content writing for search. Use when creating or editing
+  public-facing pages under editor/app/(www), writing blog posts, authoring
+  docs, or reviewing SEO-related metadata. Trigger phrases: "SEO", "metadata",
+  "sitemap", "open graph", "og image", "meta tags", "search ranking",
+  "structured data", "JSON-LD".
+---
+
+# SEO
+
+Guidelines for search engine optimization across the Grida project -- Next.js
+pages, blog posts, and documentation.
+
+## Scope
+
+| Surface                   | Location                          | Framework  |
+| ------------------------- | --------------------------------- | ---------- |
+| Marketing / product pages | `editor/app/(www)/`               | Next.js 16 |
+| Blog                      | `docs/blog/` or `apps/docs/blog/` | Docusaurus |
+| Documentation             | `docs/**`                         | Docusaurus |
+| Sitemaps                  | `editor/app/sitemap.ts` + others  | Next.js    |
+
+## Next.js Pages (`editor/app/(www)`)
+
+### Metadata
+
+Every public page must export metadata (static or dynamic).
+
+```tsx
+// Static
+export const metadata: Metadata = {
+  title: "Page Title — Grida",
+  description: "Concise, keyword-rich description under 160 chars.",
+  keywords: ["relevant", "keywords"],
+  openGraph: {
+    title: "Page Title — Grida",
+    description: "Same or tailored OG description.",
+    images: ["/og/page-name.png"],
+  },
+  twitter: {
+    card: "summary_large_image",
+  },
+};
+
+// Dynamic
+export async function generateMetadata({ params }): Promise<Metadata> { ... }
+```
+
+**Rules:**
+
+- `title` should end with ` — Grida` (or use a `title.template` in the root layout).
+- `description` must be unique per page, under 160 characters, and include primary keywords.
+- Every page with OG images must set `metadataBase` or use absolute URLs so crawlers resolve them correctly.
+- OG images must return **200** without auth and use correct `Content-Type`.
+
+### `metadataBase`
+
+When `openGraph.images` uses relative paths, the layout or page must set:
+
+```tsx
+metadataBase: new URL("https://grida.co"),
+```
+
+Without this, crawlers may see broken OG URLs.
+
+### Sitemap
+
+Public pages belong in `editor/app/sitemap.ts`.
+
+```ts
+{
+  url: "https://grida.co/new-page",
+  changeFrequency: "monthly",
+  priority: 0.5,
+}
+```
+
+- Use `priority: 1` only for the homepage.
+- Match `changeFrequency` to actual update cadence.
+- When adding a new public page, always add a sitemap entry.
+
+### Google Image Search
+
+Next.js `<Image>` renders optimized `/_next/image?url=...` URLs that Google
+Images indexes unreliably.
+
+**For pages where images are the search target** (brand assets, logos, press
+kits):
+
+- Use `<Image unoptimized>` or plain `<img>` so the rendered `src` is a
+  stable, direct public URL (e.g. `/brand/grida-symbol-240.png`).
+- Include descriptive `alt` text with "Grida" + asset name + format.
+
+**Checklist for image assets:**
+
+- Returns **200** without cookies; not blocked for `Googlebot-Image`
+- Correct `Content-Type` (`image/png`, `image/svg+xml`, etc.)
+- Not blocked by `robots.txt` or `X-Robots-Tag`
+- `/_next/image` endpoint itself is not blocked (other pages still use it)
+- Stable, clean URLs under `public/` -- no query-string-only canonicals
+
+### Structured Data (JSON-LD)
+
+Use JSON-LD for rich results where applicable:
+
+- **Product pages**: `Product` or `SoftwareApplication`
+- **Blog posts**: `BlogPosting` or `Article`
+- **FAQ sections**: `FAQPage`
+- **Brand/org**: `Organization`
+
+Embed via a `<script type="application/ld+json">` in the page or layout.
+
+## Documentation (`docs/**`)
+
+Docs are built with Docusaurus and published at `grida.co/docs`.
+
+### Frontmatter
+
+Every doc page should have:
+
+```md
+---
+title: Feature Name
+description: One-line summary for search snippets (under 160 chars).
+keywords: [keyword1, keyword2]
+---
+```
+
+- `title` becomes the `<title>` tag and H1.
+- `description` becomes the meta description.
+- `keywords` help Docusaurus generate meta tags.
+- Add `slug:` only when the default URL path is wrong.
+
+### Headings
+
+- Use a single `# H1` matching the `title` frontmatter.
+- Use `##` for sections -- these generate anchor links and appear in the TOC.
+- Headings should be descriptive and keyword-aware (search engines weight them).
+
+### Internal Linking
+
+- Link related docs to each other -- it helps search engines discover pages
+  and improves user navigation.
+- Use relative paths within `/docs`.
+- For editor pages, use universal routing: `https://grida.co/_/<path>`.
+
+## Blog Posts
+
+### Title
+
+- Clear, specific, keyword-forward. Front-load the most important words.
+- Avoid clickbait or vague titles.
+- Good: "How Grida Canvas Renders 10K Nodes at 60 FPS"
+- Bad: "Exciting Updates!"
+
+### Meta Description
+
+- Write a custom `description` in frontmatter (under 160 chars).
+- Summarize the post's value proposition -- what will the reader learn?
+
+### Content Structure
+
+- Open with a clear summary paragraph (search engines often use this as
+  the snippet).
+- Use `##` headings that include relevant keywords.
+- Break content into scannable sections.
+- Include alt text on all images.
+
+### Social Sharing
+
+- Set `image:` in frontmatter for the OG image.
+- OG image should be 1200x630 px for optimal display.
+- Ensure the image URL is absolute and publicly accessible.
+
+## Crawl Control
+
+### `robots.txt`
+
+Located at `editor/app/robots.txt`. Currently allows everything except
+`/private/`. When adding new route groups that should not be indexed (e.g.
+internal tools, preview embeds, auth flows), add a `Disallow:` rule here.
+
+### Canonical URLs & Tenant Domains
+
+Grida serves `grida.co` (main) and `[tenant].grida.site` (tenant sites).
+Tenant-rendered pages must **not** compete with `grida.co` pages in search.
+
+- Public marketing pages under `(www)` should set canonical to
+  `https://grida.co/...`.
+- Tenant pages (`(tenant)`) should either set `noindex` or canonical to their
+  own `*.grida.site` domain -- never to `grida.co`.
+- Preview / embed pages (`(preview)`) should be `noindex, nofollow`.
+
+### Known Gaps
+
+- The `(www)` root layout (`editor/app/(www)/layout.tsx`) does not set
+  `metadataBase`. Pages with relative OG images should set it themselves
+  (see `metadataBase` section above) or it should be added to the root layout.
+
+## General SEO Writing Rules
+
+These apply across all surfaces (pages, docs, blog):
+
+1. **One primary keyword per page.** Include it in the title, first paragraph,
+   and at least one heading.
+2. **Write for humans first.** Natural, helpful content ranks better than
+   keyword-stuffed text.
+3. **Unique content.** Every page must have a unique title and description.
+   Duplicate meta across pages hurts rankings.
+4. **Descriptive link text.** Use meaningful anchor text, not "click here."
+5. **Alt text on images.** Describe what the image shows; include keywords
+   when natural.
+6. **URL structure.** Keep URLs short, lowercase, hyphenated, and descriptive.
+   Avoid IDs or query params in public URLs.
+7. **Mobile-first.** Ensure pages render well on mobile -- Google uses
+   mobile-first indexing.
+8. **Performance.** Page speed is a ranking factor. Prefer static/SSG pages
+   for marketing content. Lazy-load below-fold images.
+
+## Verification Checklist
+
+Before shipping a public page or post:
+
+- [ ] `title` and `description` set in metadata / frontmatter
+- [ ] `description` is under 160 characters and unique
+- [ ] OG image set, absolute URL, returns 200 without auth
+- [ ] Page added to sitemap (Next.js pages)
+- [ ] Headings use keywords and follow hierarchy (no skipped levels)
+- [ ] All images have `alt` text
+- [ ] No duplicate meta with other pages
+- [ ] Page is mobile-friendly
+- [ ] Canonical URL is correct (no competing duplicates)
diff --git a/editor/app/(www)/SEO.md b/editor/app/(www)/SEO.md
deleted file mode 100644
index 413945ed3c..0000000000
--- a/editor/app/(www)/SEO.md
+++ /dev/null
@@ -1,48 +0,0 @@
-### SEO gotchas for `editor/app/(www)` (not the basics)
-
-This doc exists to record **easy-to-miss stack-specific issues** and **our preferred tricks**.
-
----
-
-### Google Image Search gotcha: Next.js `<Image>` hides the “real” image URL
-
-**Symptom:** the page ranks (e.g. “Grida logo png”), but **Google Images doesn’t surface the logo assets** from that page.
-
-**Cause:** Next.js `<Image>` commonly renders an optimized `src` like:
-
-- `/_next/image?url=%2Fbrand%2Fgrida-symbol-240.png&w=...&q=...`
-
-Google Images indexes the asset more reliably when the rendered `img src` is a **stable public URL** such as `/brand/...png` rather than only an optimized proxy URL.
-
-**Our trick (for brand/press/asset pages):** ensure the rendered `img src` is a **direct, stable, public asset URL**.
-
-- **Preferred**: keep `<Image>`, but add `unoptimized` for SEO-critical assets.
-- **Alternative**: use a plain `<img>` when you want to be maximally explicit.
-
-**Apply to:** pages where the images themselves are the query target (“logo png”, brand assets, downloads).
-
-**Sanity checklist:**
-
-- **Public**: returns **200** without cookies; not blocked for `Googlebot-Image`
-- **Correct `Content-Type`**: `image/png`, `image/svg+xml`, etc.
-- **Not blocked by robots**: `robots.txt` and `X-Robots-Tag`
-- **`/_next/image` is fetchable**: not blocked by robots, authentication, or middleware (other pages may still rely on the optimizer)
-- **Stable, clean URLs**: prefer `public/` paths like `/brand/...` (avoid query-string-only canonical asset URLs)
-- **Alt text**: include “Grida” + asset name + format (e.g. “Grida wordmark logo (PNG)”)
-
-**Reminder:** JSON-LD / Open Graph help, but they **don’t replace** crawlable image URLs in the HTML for Image Search.
-
----
-
-### Image search boost: sitemap + canonical URL
-
-- **Image sitemap**: include brand PNGs in the image sitemap (one entry per canonical asset URL).
-- **Canonical**: keep a single, clean canonical `/brand` URL (avoid duplicates, tracking params, and competing alternates).
-
----
-
-### Next.js metadata: relative OG/Twitter images need `metadataBase`
-
-OG/Twitter images must resolve to **absolute public URLs** that return **200** without auth.
-
-If `openGraph.images` uses relative paths in Next.js metadata, `metadataBase` is required so crawlers see correct absolute URLs.