@typeroll/mcp-server 0.41.0 → 0.42.0

package/AGENTS.md

@@ -111,8 +111,9 @@ maps to one HTTP endpoint; the actual logic runs in the customer's portal
   API key. Read the installation before writing so you use the exact keys from
   `manifest.config_schema`. This is the supported automation path for public
   content such as consent copy, link text, and policy URLs; masked secrets are
-  preserved when omitted. A successful update returns
-  `redeploy_required: true`; review the result, then deploy explicitly.
+  preserved when omitted. The update queues a production deploy by default;
+  pass `deploy: false` only when batching several changes and deploy once after
+  the final update.
 
 - **Page templates.** A `PageTemplate` is a Block[] tree that wraps a
   page's body. The template contains exactly one block of type
@@ -994,7 +995,7 @@ preview.
 | **Forms** | `list_forms`, `read_form`, `create_form`, `update_form`, `delete_form`, `list_form_submissions`, `delete_form_submission` (removes one submission — e.g. cleaning up a test entry; `delete_form` with `delete_submissions` is the bulk path). **Steps (form/* block trees) are the ONLY stored model**: pass `steps` for funnels, or `fields` for simple forms — the server converts a flat field list to a single static step. Place with a `core/form` block on block-mode pages or `<x-form id="…" />` in HTML mode. Both expand server-side to the same complete signed shell and initial state. Email/webhook actions are admin-only in the portal and excluded from agent reads/writes. |
 | **Settings** | `update_site_settings` (whitelist) |
 | **Core modules** | `list_apps`, `read_app`, `update_app` (legacy API name; admin; schema-driven config, masked secrets, redeploy when `affects_build` is true) |
-| **Extension installations** | `list_extension_installations`, `read_extension_installation`, `update_extension_installation_config` (admin; schema-driven config, masked secrets preserved, redeploy after frontend-facing changes) |
+| **Extension installations** | `list_extension_installations`, `read_extension_installation`, `update_extension_installation_config` (admin; schema-driven config, masked secrets preserved, production deploy queued by default) |
 | **Analytics attribution** | `read_funnel_attribution`, `update_funnel_attribution` (specialized Analytics module tools; admin; redeploy after changes) |
 | **Search + bulk** | `search_pages`, `check_internal_links`, `bulk_replace_text`. The link check is database-driven. Bulk replace defaults to pages but can target partials, collection items or all resources, always dry-run first. |
 | **Branches** | `create_branch`, `read_version`, `delete_branch`, `merge_branch` |

package/README.md

@@ -6,9 +6,10 @@ site through the same tool surface a human agency would use: read and
 write pages, partials, collections, media, redirects, versions; trigger
 deploys; mint preview links.
 
-The server is a **thin transport adapter** — every tool wraps one HTTP
-endpoint of the Typeroll REST API. Auth happens at the API layer with a
-site- or org-scoped key; the MCP just carries the bearer through.
+The server is a **thin transport adapter** — tools call the Typeroll REST API,
+with a few workflow tools composing consecutive API calls such as config plus
+deploy. Auth happens at the API layer with a site- or org-scoped key; the MCP
+just carries the bearer through.
 
 ## Two ways to connect
 
@@ -98,12 +99,16 @@ APIs as the portal:
 typeroll extension validate
 typeroll extension push --draft
 typeroll extension install --site test-site --config local-extension-config.json
+typeroll extension configure --site test-site --installation install-abc \
+  --config local-extension-config.json
 typeroll extension promote 1.0.0
 ```
 
 The manifest defaults to `typeroll-extension.json`; use `--manifest` to select
 another file. Local validation is a fast preflight. The portal always performs
 the complete schema, compatibility, origin and asset-hash validation.
+`extension configure` queues a production deploy by default; pass
+`--no-deploy` only when batching updates and deploy once afterwards.
 
 ## What the agent should read first
 
@@ -177,7 +182,8 @@ the full reference + concrete operation recipes.
   manifest-defined installation config through the API key with
   `update_extension_installation_config`; omitted and masked secrets are
   preserved. Use this for frontend config such as consent copy and policy
-  links, then redeploy.
+  links. It queues a production deploy by default; pass `deploy: false` only
+  when batching changes and deploy once afterwards.
 - **Settings** — read + patch, including `scripts_head` /
   `scripts_body_end` / `custom_css` (trusted because the caller holds an
   API key; the in-portal chat AI does NOT get these).

package/dist/bundled-content.js

@@ -18,13 +18,13 @@ export const BUNDLED_SKILLS = {
     "tr-migrate-astro": "---\nname: tr-migrate-astro\ndescription: Use when the user wants to migrate an Astro site — particularly an Astro Content Collections-backed site — to Typeroll. Walks `src/content/<collection>/*.md(x)`, lifts frontmatter into Typeroll collection schemas, converts markdown bodies into richtext fields, batch-imports items, then maps `src/pages/*` into Typeroll pages/partials. Triggers on \"migrate an Astro site\", \"import from src/content\", \"convert content collections\", or when the user names a local Astro repo as the source.\n---\n\n# Migrate an Astro site to Typeroll\n\n> **The buffer model (draft writes).** Every content write in this recipe\n> (pages, blocks, partials, collection items) lands in an unsaved per-doc\n> DRAFT — deploys and plain previews only see SAVED content. For recipe-style\n> build work, pass `save: true` on write calls (the work is pre-approved by\n> the task itself), or run `commit_working_copy` per doc before any\n> `trigger_deploy`. Preview your drafts with `include_working_copy: true`.\n\n\nAstro's [Content Collections](https://docs.astro.build/en/guides/content-collections/) and Typeroll's `Collection` + `CollectionItem` map one-to-one. A collection in Astro is a directory of frontmatter-bearing files under `src/content/<name>/`; in Typeroll it's a schema + items doc set under `organizations/{org}/sites/{site}/collections/{name}/`. The Astro schema (`zod.object(...)` in `src/content/config.ts`) is your field list. The frontmatter values are field values. The markdown bodies are richtext fields.\n\nThis skill walks the migration from a checked-out Astro repo on the user's machine to a target Typeroll site. You run it locally — the source repo is on disk, the MCP just receives the final shape.\n\n## Preconditions\n\n- The Astro repo is checked out locally and `npm install`d (so we can read `src/content/config.ts` to extract the schemas).\n- `@typeroll/mcp-server` configured with a valid `TYPEROLL_API_KEY` pointing at the target site (or org-scoped key + a `site_id` argument per call).\n- Target Typeroll site exists. **Empty starter site is best.** If non-empty, treat existing pages and collections as off-limits unless the user explicitly says otherwise.\n- The Astro design / theme is **not** being migrated — Typeroll has its own design layer. We migrate content; the user re-skins on the Typeroll side.\n\n## Recipe\n\n### 1. Map the source\n\nFrom the Astro repo root:\n\n```bash\nls src/content/                  # which collections exist?\ncat src/content/config.ts        # collection schemas (zod)\nls src/pages/                    # standalone pages\nls public/                       # static assets and images\n```\n\nBuild a working manifest:\n\n| Astro source | Typeroll target |\n|---|---|\n| `src/content/blog/*.md` | Collection `blog` with items |\n| `src/content/projects/*.mdx` | Collection `projects` with items |\n| `src/pages/about.astro` | Page with slug `about` |\n| `src/pages/services/[slug].astro` | If dynamic from a collection → that collection's `route_template`. If genuinely per-page → individual Typeroll pages. |\n| `public/og/*.png`, `public/images/*` | Upload to Typeroll media via `upload_media_from_url` (after staging them on a temporary public URL) or via local upload if the MCP supports it. |\n| `src/layouts/*.astro` | Header / footer / shared chunks → Typeroll partials. The rest of the layout is the site design, owned by the target site. |\n| `src/components/*.astro` | Either become partials (if reused across pages) or get inlined into the page that uses them. |\n\n### 2. Learn the target's design (don't skip)\n\n```\nget_site\nread_site_settings           # colours, fonts, voice\nlist_partials\nread_partial partial_id=\"header\"\nlist_pages limit=5\n```\n\nSame rule as in `tr-migrate-wp`: you're moving content into the *target's* visual language, not preserving the source's. Note the existing fonts, colour vars, header structure.\n\n### 3. Translate one collection schema\n\nPick the most representative collection first (usually `blog`). Read its zod schema:\n\n```ts\n// src/content/config.ts\nconst blog = defineCollection({\n  type: 'content',\n  schema: z.object({\n    title: z.string(),\n    description: z.string(),\n    pubDate: z.coerce.date(),\n    updatedDate: z.coerce.date().optional(),\n    heroImage: z.string().optional(),\n    author: z.string().default('Editorial'),\n    tags: z.array(z.string()).default([]),\n    draft: z.boolean().default(false),\n  }),\n});\n```\n\nMap zod types to Typeroll field types:\n\n| Astro zod | Typeroll field type |\n|---|---|\n| `z.string()` | `text` (or `textarea` if it's a description/excerpt — judge by typical length) |\n| `z.string().long()` / a description field | `textarea` |\n| `z.coerce.date()` / `z.date()` | `date` |\n| `z.number()` | `number` |\n| `z.boolean()` | `boolean` |\n| `z.string()` with image path / `image()` helper | `image` |\n| `z.array(z.string())` | `text` (comma-joined) or `tags` if you have a tags field type |\n| `z.enum([...])` | `text` with a comment about the allowed values; the model writes the listing logic |\n| `z.object({...})` (nested) | Flatten into prefixed fields, or pre-render into a `*_html` field (see `tr-collection-template`) |\n| Markdown body | `body: richtext` (the markdown content of the file, converted to HTML — see §4) |\n\nCreate the collection with the design template baked in (this is the new pattern — read `tr-blog` if you haven't yet):\n\n```\ncreate_collection {\n  \"name\": \"blog\",\n  \"label_singular\": \"Article\",\n  \"label_plural\": \"Articles\",\n  \"slug_field\": \"slug\",\n  \"sort_field\": \"date\",\n  \"sort_dir\": \"desc\",\n  \"route_template\": \"/blog/{slug}\",\n  \"fields\": [\n    {\"name\":\"title\",       \"type\":\"text\",     \"required\":true},\n    {\"name\":\"slug\",        \"type\":\"text\",     \"required\":true},\n    {\"name\":\"description\", \"type\":\"textarea\"},\n    {\"name\":\"date\",        \"type\":\"date\",     \"required\":true},\n    {\"name\":\"updated_date\",\"type\":\"date\"},\n    {\"name\":\"hero_image\",  \"type\":\"image\"},\n    {\"name\":\"author\",      \"type\":\"text\"},\n    {\"name\":\"tags\",        \"type\":\"text\"},\n    {\"name\":\"body\",        \"type\":\"richtext\"}\n  ],\n  \"item_template_html\": \"<article class=\\\"post\\\">...</article>\"\n}\n```\n\nField-name rules: ASCII, lowercase, `[a-z][a-z0-9_-]*`. The Astro source `pubDate` → Typeroll `date`. `updatedDate` → `updated_date` (snake_case is fine; camelCase isn't).\n\n### 4. Convert markdown bodies → richtext\n\nAstro stores the markdown body as the file content after the `---` frontmatter fence. Typeroll's `body: richtext` wants HTML. Pick one of:\n\n**Option A — use Astro's own markdown renderer (preferred when the repo already builds):**\n\n```js\nimport { unified } from 'unified';\nimport remarkParse from 'remark-parse';\nimport remarkRehype from 'remark-rehype';\nimport rehypeStringify from 'rehype-stringify';\n\nconst md2html = async (md) => {\n  const file = await unified()\n    .use(remarkParse)\n    .use(remarkRehype, { allowDangerousHtml: true })\n    .use(rehypeStringify, { allowDangerousHtml: true })\n    .process(md);\n  return String(file);\n};\n```\n\nRun this on the body of every collection file. Image references inside the markdown (`![alt](./hero.png)`) need their `src` rewritten to the uploaded Typeroll CDN URLs after step 6.\n\n**Option B — convert ad-hoc:** `marked`, `markdown-it`, or any other parser. Just stay consistent across files so the HTML output looks uniform.\n\nTyperoll's sanitiser will strip `<script>` and event handlers from the output regardless. If the markdown carried embedded raw HTML you want to keep (iframes for video, etc.), check the sanitiser config in `packages/site-template/src/lib/sanitize.ts` for the whitelist.\n\n### 5. Resolve and upload images\n\nFor every image referenced by an item (`heroImage`, images inside the markdown body):\n\n1. Stage the local file at a temporary public URL (or use a local-upload MCP tool if one is configured).\n2. `upload_media_from_url url=<staged-url> alt=<from frontmatter or filename>` — record the returned CDN URL.\n3. Substitute the CDN URL into both the `hero_image` field value AND the markdown-converted HTML body (regex replace `src=` references).\n\nKeep a local map (`./astro-migration-state.json`):\n\n```json\n{\n  \"media\": {\n    \"./hero.png\": \"https://cdn.typeroll.com/<orgId>/<siteId>/abc123.png\"\n  }\n}\n```\n\nSo a partial run is resumable and you don't re-upload the same image twice.\n\n### 6. Batch-import items\n\nFor each file in `src/content/<collection>/`:\n\n```js\nimport fs from 'node:fs';\nimport path from 'node:path';\nimport matter from 'gray-matter';\n\nconst files = fs.readdirSync('src/content/blog').filter(f => /\\.mdx?$/.test(f));\nfor (const file of files) {\n  const raw = fs.readFileSync(`src/content/blog/${file}`, 'utf8');\n  const { data, content } = matter(raw);\n  const slug = file.replace(/\\.mdx?$/, '');\n  const body = await md2html(content);\n\n  // create_collection_item via the MCP\n  await mcp.callTool('create_collection_item', {\n    collection: 'blog',\n    status: data.draft ? 'draft' : 'published',\n    fields: {\n      title:        data.title,\n      slug,\n      description:  data.description,\n      date:         data.pubDate ? new Date(data.pubDate).toISOString().slice(0, 10) : null,\n      updated_date: data.updatedDate ? new Date(data.updatedDate).toISOString().slice(0, 10) : null,\n      hero_image:   mediaMap[data.heroImage] || data.heroImage,\n      author:       data.author,\n      tags:         (data.tags || []).join(', '),\n      body,\n    },\n  });\n}\n```\n\nRate-limit awareness: spawn at most 5 parallel `create_collection_item` calls; the API caps at ~60 writes/minute and responds with 429 + `Retry-After` if exceeded.\n\nSet `status: 'draft'` (or honour Astro's `draft: true` frontmatter) for items the user should review before publishing. Only published items get static pages.\n\n### 7. Build the listing page\n\n```\ncreate_page title=\"Blog\" slug=\"blog\" status=\"published\" content_mode=\"html\"\n  html_content=\"<section><h1>Blog</h1>\n  <!-- typeroll:listing:blog -->\n  <!-- /typeroll:listing:blog -->\n</section>\"\n\nregenerate_collection_listing\n  collection=\"blog\"\n  page_id=\"blog\"\n  item_template=\"<article class=\\\"blog-card\\\"><a href=\\\"{{url}}\\\"><h2>{{title}}</h2><p>{{description}}</p><time>{{date}}</time></a></article>\"\n  wrap_open=\"<div class=\\\"blog-grid\\\">\"\n  wrap_close=\"</div>\"\n```\n\nSee `tr-blog` for full styling and post-import update flow.\n\n### 8. Translate standalone pages\n\nFor each non-dynamic `.astro` page under `src/pages/`:\n\n```\nread_partial partial_id=\"header\"          # learn target's nav style\n# Re-skin the page content using target site's CSS variables and partials.\ncreate_page title=\"About\" slug=\"about\" status=\"draft\" content_mode=\"html\"\n  html_content=\"...\"\n```\n\nDefault to **draft** — the user reviews each page before publishing.\n\nFor dynamic Astro pages (`[slug].astro` that consume a collection), you're already done: the matching Typeroll collection's `route_template` produces the same URLs at build time.\n\n### 9. Add redirects for URL changes\n\nIf Astro's slugs differed from what you derived for Typeroll (e.g. Astro had `/posts/my-article` but you want `/blog/my-article`), bulk-add redirects:\n\n```\nadd_redirect from=\"/posts/my-article\" to=\"/blog/my-article\" status=301\n```\n\nOr build a redirect map from the `astro-migration-state.json` and apply it in one pass.\n\n### 10. Deploy\n\n```\ntrigger_deploy\nget_deploy_status job_id=<id>\n```\n\nBrowse the resulting site, compare against the Astro source, surface anything that drifted.\n\n## Astro-specific gotchas\n\n- **`.mdx` files with custom components.** Components inside MDX (`<MyCustom prop=\"...\" />`) will not render in Typeroll because the component definitions don't migrate. Two options: (1) replace each component with its rendered HTML output (run the Astro build, scrape the rendered HTML, use that as `body`); (2) if the component is a reusable visual element used across many items, factor it into a Typeroll partial and replace MDX usages with `<x-include name=\"...\" />` calls. See `tr-page-template` for the partial-include pattern.\n- **`src/content/config.ts` typed `image()` helper.** Astro resolves `image()` fields at build time to optimised assets. After migration the Typeroll `hero_image` field carries the original source URL — re-upload via step 5 to get it onto the Typeroll CDN.\n- **`getCollection` filters at runtime.** If `src/pages/blog/index.astro` does `getCollection('blog', ({ data }) => !data.draft)`, Typeroll handles this via `status` on the item. Don't translate the filter — set `status: 'draft'` on items where `data.draft === true`.\n- **Per-tag pages (`/blog/tag/[tag].astro`).** Typeroll doesn't auto-generate these. Either (a) drop tag pages and use a client-side filter in the blog listing JS, (b) generate them by enumerating unique tags and calling `create_page` per tag with a server-side-pre-filtered listing. (a) is preferable for ≤dozens of tags.\n- **`rehype-pretty-code` / shiki / fenced code with syntax highlighting.** The HTML output of these contains inline styles that the Typeroll sanitiser preserves. The colours match the Astro site's theme at import time; redoing the target design later means re-running the highlighter against the same markdown source. Keep a copy of the raw markdown if you might.\n- **`og:image` per page from `astro-og-image` style plugins.** Typeroll has its own SEO surface (`seo_title`, `seo_description`, `seo_og_image`). Set them explicitly on each page during step 8.\n- **i18n via `src/content/<lang>/<collection>/`.** Typeroll's site-level `default_language` + per-page `language` field cover this (capabilities: `supports_language_per_page: true`). Map each language directory to per-item `language: 'sv-SE' | 'en-US' | …` instead of separate collections.\n\n## Mixing imported + new content\n\nHalf-migrate, leave drafts, let the user inspect, iterate. You can:\n\n- Import only collections, skip standalone pages, let the user rebuild those from scratch.\n- Import everything as `status: 'draft'`, treat publishing as a per-item human review pass.\n- Mix sources: pull article bodies from Astro markdown, use Claude to generate fresh excerpts/SEO metadata before writing the item.\n\nKeep the local `astro-migration-state.json` honest — it's the only way to make a partial run resumable when the API rate-limits or a markdown parser trips on an edge case file.\n",
     "tr-migrate-multisite": "---\nname: tr-migrate-multisite\ndescription: Use when migrating SEVERAL sites at once — a WordPress multisite network, a group of country/language sites on different top-level domains, or any batch of related sites moving to Typeroll together. Triggers on \"multisite\", \"network of sites\", \"10 sites\", \"our .se/.de/.co.uk sites\", \"language versions\", \"migrate all our sites\". For a single WordPress site use tr-migrate-wp; for a single non-WP source use tr-import-url.\n---\n\n# Migrate a multisite / multi-domain family to Typeroll\n\n> **The buffer model (draft writes).** Every content write in this recipe\n> (pages, blocks, partials, collection items) lands in an unsaved per-doc\n> DRAFT — deploys and plain previews only see SAVED content. For recipe-style\n> build work, pass `save: true` on write calls (the work is pre-approved by\n> the task itself), or run `commit_working_copy` per doc before any\n> `trigger_deploy`. Preview your drafts with `include_working_copy: true`.\n\n## The first decision: one site or many?\n\n**One Typeroll site owns exactly one domain** (plus its apex/www sibling).\nSo:\n\n| The old family looks like | Build it as |\n|---|---|\n| `example.se`, `example.de`, `example.co.uk` — separate domains | **N Typeroll sites**, one per domain |\n| `example.com/se/`, `example.com/de/` — one domain, language folders | **One site**, using `path` on each page (`/se/om-oss`) |\n| A WP multisite on subdomains that the customer wants to consolidate onto one domain | **One site** + redirects from every old subdomain |\n\nConfirm this with the user before creating anything — it's the one decision\nthat is expensive to reverse (domains, deploys, and analytics all hang off\nit). The rest of this recipe assumes the common case: **N sites, one per\ndomain**, sharing a design.\n\nRequires an **org-scoped API key** (`create_site` needs it, and one key\nreaching every site is the whole point here). With MCP you pass `site_id`\nper call; over stdio you'll be re-pointing `TYPEROLL_SITE_ID` per site.\n\n## Phase 0 — Plan the batch (do this once, in writing)\n\nProduce a table and get the user to confirm it before touching the platform:\n\n| Old URL | Language | New Typeroll site | Domain | Notes |\n|---|---|---|---|---|\n\n**Check platform readiness before the plan is even agreed:**\n\n```\nget_migration_readiness site_id=<any existing site>\n```\n\nThen once per market, with that market's own source:\n\n```\nget_migration_readiness site_id=<de-site> source_url=\"https://example.de\"\n```\n\nThe blockers (media storage, hosting credentials) are **platform-level, not\nper-site** — if they fail for one site they fail for all ten, and finding out\nafter building three sites means redoing three sites' worth of image work.\nGet them fixed before Phase 1, and re-run the check per site once the sites\nexist (the warnings — verification URL, form email, design reference — are\nper-site).\n\nAsk explicitly:\n- Which site is the **design reference**? Build that one properly first.\n- Are the sites **translations of each other** (same page structure) or\n  independent? This decides whether hreflang clusters are mechanical or\n  hand-mapped.\n- Any domains being **retired or merged**? Those need redirects at the DNS\n  level, not just inside a site.\n\n## Phase 1 — Inventory EVERY old site, before building anything\n\nDo this for all sites up front. It's cheap, it's the only artefact that tells\nyou when you're done, and it stops you discovering an untouched 400-URL blog\nin week three.\n\nFor each source site, create the target site first (the inventory lives on\nit):\n\n```\ncreate_site name=\"Example DE\" domain=\"example.de\"\n```\n\nThen walk the source and post what you find:\n\n```\nfetch https://example.de/sitemap.xml            # follow sitemap-index children\nfetch https://example.de/wp-json/wp/v2/pages?per_page=100   # walk X-WP-TotalPages\n```\n\n```\nadd_migration_urls site_id=<de-site> source_origin=\"https://example.de\" source=\"sitemap\" urls=[\n  { url: \"https://example.de/ueber-uns\" },\n  { url: \"https://example.de/kontakt\" },\n  …\n]\n```\n\n**`source_origin` is not optional in a multisite job.** It rejects URLs from\nanother origin, which is the guard that stops domain B's `/kontakt` landing\nin domain A's inventory — where it would silently read as \"covered\" because\ndomain A happens to have a `/kontakt` too.\n\nAdd every source you have, each with its own label — they merge per URL:\n- `source=\"sitemap\"` — the sitemap(s)\n- `source=\"rest\"` — WP REST pages/posts/custom types\n- `source=\"gsc\"` — a Search Console export, **with `gsc_clicks`**. This is\n  what makes the coverage report prioritise itself: the 12 URLs carrying all\n  the traffic sort to the top.\n- `source=\"crawl\"` — anything you found by following internal links\n\nMark the obvious throwaways immediately, so the work list is real:\n\n```\nadd_migration_urls urls=[{ url: \"/wp-admin\", excluded: true }, { url: \"/tag/nyheter\", excluded: true }]\n```\n\nRead it back and note the starting number:\n\n```\nlist_migration_urls site_id=<de-site> status=\"unhandled\" limit=50\n```\n\n## Phase 2 — Build the reference site\n\nFollow `tr-migrate-wp` (or `tr-import-url`) for the ONE reference site:\ndesign, header/footer partials, page templates, block types. Get the user to\napprove it before replicating — every fix you make after this point costs\nN times as much.\n\n## Phase 3 — Replicate the design to the other sites\n\nDesign travels as a block-type package, not by hand:\n\n```\nexport_block_types site_id=<reference>                 # → .tcblocks JSON\nimport_block_types site_id=<other> package=<that JSON>\n```\n\nThen per site:\n- `read_site_settings` on the reference → `update_site_settings` on the\n  target with the same colors/fonts (translate `site_name`, `tagline`,\n  contact details — those are per-market, not shared).\n- Recreate header/footer partials with translated nav labels.\n- Page templates: rebuild with `add_block target={kind:'template', id:…}`.\n\nSet the language per site — it drives `<html lang>`, `og:locale` and\nalt-text generation:\n\n```\nupdate_site site_id=<de-site> language=\"de\"\n```\n\n## Phase 4 — Migrate content, preserving paths\n\nPer site, per URL, follow `tr-migrate-wp` §3. Two rules that matter more here\nthan in a single-site migration:\n\n1. **Preserve the path verbatim** unless there's a reason not to. Use\n   `path` for anything nested: `create_page title=\"Über uns\" slug=\"ueber-uns\"\n   path=\"/ueber-uns\"`. A preserved path needs no redirect and loses nothing.\n2. **Rewrite internal links to the NEW paths.** Imported HTML is full of\n   absolute links to the old domain. Sweep them per site:\n\n```\nbulk_replace_text site_id=<de-site> find=\"https://example.de/\" replace=\"/\" dry_run=true\n```\n\nCheck the dry-run count against what you expect before running it for real.\nCross-domain links between sister sites stay absolute — only the site's own\ndomain becomes relative.\n\n### Media: per site, not shared\n\nEvery Typeroll site has its own media library, so a shared asset (the group\nlogo, a product shot used in all markets) is uploaded once **per site** and\ngets a different CDN URL in each. That's correct — the sites are independent\nand one market's deploy must not depend on another's assets — but it means:\n\n- Don't try to reuse a `cdn_url` from site A inside site B's HTML. It will\n  render, and it will break the day site A is deleted or moved.\n- Do write alt text per market, in that market's language:\n  `update_media media_id=… alt_text=\"…\"`. The alt text is content, not\n  metadata, and a Swedish alt on a German page is a real accessibility defect.\n\nImages referenced only from a stylesheet or from unrendered page-builder JSON\nare NOT found by an HTML scan. Spot-check the hero/background images of the\ntop pages in the preview before you call a site done.\n\n### Forms are NOT migrated — plan to rebuild them\n\nThe HTML cleaner strips `<form>`, `<input>`, `<select>` and `<button>`\nentirely, on purpose: a Contact Form 7 / Gravity / Elementor form posts to\nWordPress endpoints that no longer exist, so importing the markup would give\nyou a form that looks alive and silently drops every submission.\n\nSo, per site:\n\n```\ncreate_form name=\"Kontakt\" fields=[…]         # or steps=[…] for a funnel\nadd_block target={kind:'page', id:'kontakt'} block={type:'core/form', data:{form_id:'<id>'}}\n```\n\nThen, still per site:\n\n- **Recipient address per market** — the German enquiries rarely go to the\n  Swedish inbox. Check this explicitly; it is the single most common thing\n  to get wrong in a batch of ten.\n- **Email delivery is configured per site** by an admin in the portal\n  (Settings → Integrations), not through this API. Flag it to the user as a\n  manual step — a form that saves submissions but sends no notification looks\n  fine in testing and loses leads in production.\n- **Submit a real test through every form** after deploy, and confirm both the\n  stored submission and the notification email.\n\nCount the old site's forms during Phase 1 and put them in the plan table.\nTen sites × three forms is thirty forms, and it is the part of the job that\nnever shows up in a URL inventory.\n\n## Phase 5 — Redirects for everything you didn't preserve\n\nWork the coverage report, not your memory:\n\n```\nlist_migration_urls site_id=<de-site> status=\"unhandled\"\n```\n\nFor each entry, one of three outcomes — no fourth option:\n\n- It moved → `create_redirect from_path=\"/alte-seite\" to_path=\"/neue-seite\"`\n- It's gone on purpose → `update_migration_url url_id=… excluded=true notes=\"Old campaign LP, signed off by <name>\"`\n- It should exist and doesn't → go back and migrate it\n\nRe-read the list. `unhandled` reaching zero is the exit condition for this\nphase.\n\n**Clear the URL families with one rule each**, per site — a WP network\nmultiplies the same dead shapes across every market:\n\n```\ncreate_redirect site_id=<de-site> from_path=\"/category/*\" to_path=\"/blogg/:splat\"\ncreate_redirect site_id=<de-site> from_path=\"/tag/*\"      to_path=\"/blogg\"\ncreate_redirect site_id=<de-site> from_path=\"/2019/*\"     to_path=\"/blogg/:splat\"\n```\n\nOnly a TRAILING `*` is supported (`:splat` replays the remainder); `:name`\nmatches one segment. Pattern-covered inventory URLs count as `redirected`, so\nthe work list actually empties. A pattern that would hide a live page is\nrefused, naming the pages — narrow the prefix rather than working around it.\n\nWatch the per-market prefixes: the German site's archive base is `/kategorie/`,\nnot `/category/`. Write the rules from each site's own inventory, never by\ncopying the reference site's.\n\n## Phase 6 — Wire the hreflang cluster\n\nThis is the step that only exists because the family is multi-domain, and the\none most likely to be skipped. Each page declares its siblings on the other\ndomains; the renderer adds the page's own self-reference.\n\n```\nupdate_page site_id=<se-site> page_id=\"om-oss\" patch={ alternates: [\n  { hreflang: \"de\", href: \"https://example.de/ueber-uns\" },\n  { hreflang: \"en-GB\", href: \"https://example.co.uk/about-us\" },\n  { hreflang: \"x-default\", href: \"https://example.com/about-us\" }\n]}\n```\n\nRules the search engines actually enforce:\n- **Reciprocal.** Every page in a cluster must list every other one. Write\n  all N sides or the cluster is ignored. `batch_update_pages` is the sane way\n  to do this once you have the mapping table.\n- **One `x-default`** per cluster, pointing at the language selector or the\n  fallback market. Optional, but useful when the family doesn't cover a\n  visitor's language.\n- **Absolute URLs on the final domain** — not the `*.typeroll` fallback\n  subdomain. The cluster is what you want live after cutover, and a fallback\n  URL in there is a leak you'll be cleaning up for months.\n- Invalid entries are **rejected at write time** with the reason. If a write\n  fails, fix the tag/href — don't strip the field to make it pass.\n\nPages that have no equivalent on the other domains get no alternates at all.\nA cluster of one is meaningless markup.\n\n## Phase 7 — Verify BEFORE touching DNS\n\nDeploy each site (`trigger_deploy`), then check what it actually serves:\n\n```\nverify_migration_urls site_id=<de-site> source_origin=\"https://example.de\" check_source=true\n```\n\nThis requests every inventory URL against the site's fallback subdomain —\nthe real domain still points at the old host, which is exactly why the check\nis possible at all. Verdicts:\n\n| Verdict | Meaning | Action |\n|---|---|---|\n| `ok` | 200 at the same path | none |\n| `ok_redirect` | redirects to a 200 | none; flatten if `hops` > 1 |\n| `missing` | 404/410 | **the gap** — redirect it or migrate it |\n| `broken_redirect` | loop, or chain ending on an error | fix the rule |\n| `error` | 5xx / timeout | inconclusive, re-run |\n\n`check_source=true` also requests the OLD site, so a URL that already 404s\nupstream shows up as noise in the inventory rather than as a migration\nfailure — mark those `excluded`.\n\nNote what the check does NOT catch: it verifies that a URL *resolves*, not\nthat the page at the other end is the right content. Spot-check the top\n`gsc_clicks` URLs by eye.\n\nIterate until `missing` and `broken_redirect` are both zero **on every site**.\nThen, per site:\n\n1. `add_domain` / follow the DNS instructions the platform returns\n2. Point DNS\n3. `poll_domain` until verified → `activate_domain`\n4. Re-run `verify_migration_urls target_origin=\"https://example.de\"` against\n   the real domain, to confirm the cutover kept what the pre-check proved\n5. Submit the new sitemap in Search Console; keep the old property open for\n   a few weeks and watch the 404 report\n\n## Definition of done (per site)\n\n- [ ] `get_migration_readiness source_url=<this market's old site>` → `ready: true`, warnings reviewed\n- [ ] `list_migration_urls status=\"unhandled\"` → 0\n- [ ] `verify_migration_urls` → 0 `missing`, 0 `broken_redirect`\n- [ ] hreflang cluster written on both/all sides, absolute, final domains\n- [ ] `language` set on the site; `<html lang>` correct in the deployed HTML\n- [ ] Internal links rewritten (no lingering absolute links to the old domain)\n- [ ] Forms rebuilt, recipient address correct for THIS market, test submission sent and received\n- [ ] Media uploaded to this site's own library (no cross-site `cdn_url`), alt text in this market's language\n- [ ] Domain verified + activated; sitemap submitted\n\n## Pitfalls specific to this job\n\n- **Don't share one inventory across domains.** Inventory entries key on\n  path; two markets both have `/kontakt`. One site, one inventory.\n- **Don't build all N sites in parallel from scratch.** Build one, approve,\n  replicate. Parallel building multiplies every design mistake by N.\n- **Don't skip the parity check because coverage says 100%.** Coverage is a\n  claim about the data; parity is a measurement of the server. They disagree\n  exactly when it matters — an unpublished target page, a typo'd path, a\n  redirect chain.\n- **Don't point DNS site-by-site on a whim.** Cutting over one market at a\n  time is fine and usually wise, but the hreflang cluster spans markets: a\n  page pointing at a domain that still serves the old site is pointing at\n  content that doesn't match. Either cut over close together, or write the\n  cluster after the last market lands.\n",
     "tr-migrate-wp": "---\nname: tr-migrate-wp\ndescription: Use when the user asks to migrate a WordPress site to Typeroll, mentions wp-json, or names a WP source URL. Walks the WP REST API, rebuilds pages in the target site's design, transfers media, sets redirects, leaves everything as drafts for human review.\n---\n\n# Migrate from WordPress to Typeroll\n\n> **The buffer model (draft writes).** Every content write in this recipe\n> (pages, blocks, partials, collection items) lands in an unsaved per-doc\n> DRAFT — deploys and plain previews only see SAVED content. For recipe-style\n> build work, pass `save: true` on write calls (the work is pre-approved by\n> the task itself), or run `commit_working_copy` per doc before any\n> `trigger_deploy`. Preview your drafts with `include_working_copy: true`.\n\n\nThe platform's in-portal migration workflow is the \"managed\" path for\ncustomers who want one-click. This skill is the \"power-user\" path: you\ndo it locally, mix data sources freely, and the user (consultant /\nagency) reviews each step in their terminal.\n\n## Preconditions\n\n**Run the readiness check FIRST — before touching any content:**\n\n```\nget_migration_readiness source_url=\"https://oldsite.com\"\n```\n\nPass `source_url` — that adds the checks on the site you're migrating FROM.\nAn old host that answers 403/429 to server-side requests is a **blocker**: the\nimport would produce empty pages, or pages containing the host's block page,\nwhich reads as real content and is worse. Whether `/wp-json` answers is a\nwarning, since scraping is a real fallback (it just loses ACF/custom fields).\n\nIf `ready` is false, STOP and report the blockers to the user. Do not start\nthe import \"and fix it after\": every blocker is one whose failure is invisible\nonce the work is done, so discovering it late means redoing the expensive part.\n\n- **Media storage** — without it, every `<img>` keeps its WordPress URL. The\n  new site looks perfect and is still served images by the old host. It breaks\n  the day the customer cancels that hosting, months later, all at once.\n- **Hosting adapter** — without credentials, deploys return a job id and\n  publish nothing, while reporting success.\n\nWarnings are worth relaying but don't stop you: no verification origin (the\npre-cutover parity check can't run), no AI reconstruction key, forms without a\nnotification address, or a target site with no design to rebuild INTO.\n\nThen the ordinary preconditions:\n\n- `@typeroll/mcp-server` configured with a valid `TYPEROLL_API_KEY`.\n- The source WP site has `/wp-json` reachable (Google for \"wordpress\n  REST API disabled\" if not — common for hardened hosts).\n- The Typeroll target site exists **and already carries the design** —\n  settings, header/footer, one or two example pages. The migration rebuilds\n  old content in the NEW design; with nothing to imitate it inherits the old\n  site's look.\n- If the target already has content, you must NOT clobber it — always\n  `list_pages` first and only write to slugs that don't already exist.\n\n## Recipe\n\n### 1. Probe and inventory\n\n```\nfetch <wp-url>/wp-json                           # confirm REST is on\nfetch <wp-url>/wp-sitemap.xml or /sitemap.xml    # URL inventory\n```\n\nBuild a list of every URL you intend to migrate. WP custom post types\nneed their REST endpoint (e.g. `/wp-json/wp/v2/news?per_page=100`),\nwalking `X-WP-TotalPages` to paginate.\n\n### 2. Learn the target's design\n\n```\nget_site\nread_site_settings                               # colors, fonts, voice cues\nlist_partials                                    # header / footer / shared\nread_partial partial_id=\"header\"                 # nav structure\nlist_pages limit=5\nbatch_read_pages page_ids=[<2-3 representative ids>]   # see actual conventions\n```\n\nDon't skip this. Imposing a stranger's design on a customer's site is\nthe biggest avoidable mistake.\n\n### 3. Migrate one page at a time, draft status\n\nFor each source URL:\n\na. Fetch from WP. Prefer the helper plugin's authenticated endpoint\n   (`/wp-json/typeroll/v1/...`) if available — it bypasses\n   `show_in_rest=false` and returns ACF + builder fields. Otherwise\n   fall back to `/wp-json/wp/v2/<post-type>?slug=<slug>`.\n\nb. Clean the HTML. Strip Elementor / Gutenberg / Breakdance class\n   soup. Drop empty `<div>` and `<span>` wrappers. Keep semantic tags,\n   tables, iframes from known hosts (YouTube / Vimeo / Calendly).\n\nc. Migrate referenced images:\n   - For each `<img src>` and CSS `background-image: url()`:\n     1. Download the source image locally.\n     2. `create_upload_url filename=... content_type=...` → returns\n        `{ upload_url, cdn_url, media_id }`.\n     3. PUT the bytes to `upload_url` (curl or fetch with the same\n        content type).\n     4. Replace the `src` with `cdn_url` in the rewritten HTML.\n   - Use `update_media media_id=... alt_text=\"...\"` to set a real alt\n     text (existing WP `alt` attribute or `aria-label`; fall back to\n     filename only as a last resort).\n\nd. Reconstruct in the target's design. The cleaned HTML is rarely\n   ready to ship — typical fixes: replace WP `wp-block-*` classes\n   with the target's CSS variables; turn Elementor sections into\n   plain `<section>` with the target's spacing; fix headings so the\n   page has exactly one `<h1>`. If you're confident, batch these\n   through `bulk_replace_text` with `dry_run: true` first.\n\ne. Write the page as a draft:\n\n   ```\n   create_page title=\"...\" slug=\"<preserved-from-wp>\"\n               html_content=\"<reconstructed>\"\n               status=\"draft\" kind=\"article\" author=\"...\"\n               seo_title=\"...\" seo_description=\"...\"\n   ```\n\n   **Preserve the source URL.** WP post URLs like\n   `/2024/01/foo-bar/` go in as `slug: \"2024/01/foo-bar\"`. The\n   slug supports slashes; encode the WP permalink structure verbatim\n   when the customer wants existing links to keep working.\n\n### 4. Redirects\n\nAfter migration, every URL the agent didn't preserve verbatim needs a\nredirect:\n\n```\ncreate_redirect from_path=\"/old-services\" to_path=\"/services\"\n```\n\nWalk the inventory; for each URL: did it become a page with the same\npath? If yes, no redirect. If renamed, `create_redirect`. If\nintentionally dropped, mark it `excluded` via `update_migration_url` (the\ncustomer should sign off on every dropped URL).\n\n**Use wildcards for WordPress's URL families.** A WP site's dead URLs come in\nshapes, not as individuals — and the inventory only knows the ones it found,\nwhile the old site had more (paginated archives, feeds, attachment pages). One\npattern rule retires the whole family:\n\n| WordPress shape | Rule |\n|---|---|\n| Category archives | `from_path=\"/category/*\"` → `to_path=\"/blogg/:splat\"` (or a single landing page) |\n| Tag archives | `from_path=\"/tag/*\"` → `to_path=\"/blogg\"` |\n| Author archives | `from_path=\"/author/*\"` → `to_path=\"/om-oss\"` |\n| Date-based permalinks | `from_path=\"/2019/*\"` → `to_path=\"/blogg/:splat\"` — one rule per year |\n| Old post prefix → new | `from_path=\"/blog/:slug\"` → `to_path=\"/artiklar/:slug\"` |\n| Feeds | `from_path=\"/feed/*\"` → `to_path=\"/blogg\"` |\n\nRules:\n\n- **Trailing `*` only.** A mid-path splat (`/blog/*/comments`) is dropped\n  silently by Cloudflare — the platform refuses it at write time.\n- **`:splat`** carries the captured remainder; **`:name`** matches exactly one\n  segment and can be replayed by name.\n- **A pattern that would hide a live page is refused**, naming the pages. That\n  is the platform protecting you: redirects are applied before static files, so\n  `/blogg/*` would make every real article under `/blogg/` unreachable. Narrow\n  the prefix instead.\n- **Query-string URLs can't be matched.** WP's `/?p=123` has no path to key on;\n  those need handling at the source (or accept the loss and mark them excluded).\n- Rules are emitted most-specific-first, so a narrow rule always beats a broad\n  one — you can safely have `/blogg/recept/*` alongside `/blogg/*`.\n\nThen verify against reality before anything is cut over:\n\n```\nimport_sitemap url=\"https://old.example.com/sitemap.xml\"\n# Optional: direct Search Console query (platform service account must have property access)\nimport_gsc_performance property=\"https://old.example.com/\" months=6\n# Or paste a Search Console Pages CSV via csv=\"...\" and source_origin.\ncheck_internal_links         # database preflight before deploy\nverify_migration_urls        # after trigger_deploy; compact exceptions by default\n```\n\n### 5. Preview + review with the user\n\n```\nget_preview_link page_id=<id>                    # one URL the user can click\n```\n\nOpen in the user's browser. The preview navigates the whole site from\none mint. Iterate on feedback: pages, header, footer.\n\n### 6. Ship\n\nWhen the user signs off:\n\n```\n# Bulk-publish drafts that look right\nbatch_update_pages updates=[{page_id, patch:{status:\"published\"}}, ...]\n\n# Deploy\ntrigger_deploy\nget_deploy_status job_id=<id>    # poll\n```\n\n## Pitfalls\n\n- **Don't publish during migration.** Always import as `draft`. Even\n  if the agent is confident, the customer needs the chance to spot-check.\n- **WP slugs sometimes drift.** A post saved with slug `foo-bar` may\n  have been served at `/2024/01/foo-bar/` due to the permalink\n  structure. The full URL is what users see in Google; preserve that,\n  not the bare slug.\n- **Image bandwidth.** R2 upload is metered. Use `find_pages_matching`\n  contains=\"<old-domain>\" on already-imported content to spot images\n  that weren't transferred.\n- **WP-specific JSON-LD** (Yoast, Rank Math) is usually wrong after a\n  redesign because it references old URLs. Strip it; let Typeroll\n  emit fresh Article/Page schemas via `kind: 'article'` + `author`.\n\n## When the source isn't WordPress\n\nThe same shape applies for any source — Squarespace export, custom\nCMS, scraped HTML, CSV. Replace step 1's \"WP REST\" probe with whatever\ndiscovery the source supports, and the rest of the recipe is unchanged.\n",
-    "tr-new-site": "---\nname: tr-new-site\ndescription: Use when the user wants to create a new Typeroll site from scratch, set up the initial design, or bootstrap a blank site with working header/footer, brand colors, and a homepage. Also triggers on \"start a new site\", \"set up a site for\", or \"build a website for [company]\".\n---\n\n# Bootstrap a new Typeroll site\n\n> **The buffer model (draft writes).** Every content write in this recipe\n> (pages, blocks, partials, collection items) lands in an unsaved per-doc\n> DRAFT — deploys and plain previews only see SAVED content. For recipe-style\n> build work, pass `save: true` on write calls (the work is pre-approved by\n> the task itself), or run `commit_working_copy` per doc before any\n> `trigger_deploy`. Preview your drafts with `include_working_copy: true`.\n\n\nStart here when the site already exists as a database record (created via\nthe portal UI or API) but has no design, no header/footer, and no pages.\nThe goal is to go from blank to a working 4-page site with correct brand\nidentity in a single session.\n\n**Pages are built in block mode** — the platform default. Blocks give\nstructured, per-field editing, native full-bleed sections, responsive\nbreakpoints, and templates. HTML mode is the secondary path for\nhand-crafted one-offs and migrated content (section at the end).\n\n## Preconditions\n\n- `@typeroll/mcp-server` configured with a valid `TYPEROLL_API_KEY`.\n- The site exists (confirm with `get_site`).\n- You have the customer brief: company name, industry, 2–3 key brand colors,\n  tone of voice, and a list of initial pages.\n\n## Recipe\n\n### 1. Audit current state\n\n```\nget_site\nget_site_capabilities   # template_capabilities_version — what this deployment supports\nread_site_settings      # see what (if anything) is already configured\nlist_pages              # don't overwrite pages that already exist\nlist_partials           # check if header/footer already have content\nlist_block_types        # the per-site block palette — NEVER assume, always list\n```\n\n**Do not choose a content mode or file a platform gap before these calls\nreturn.** For every required visual/behavioral element, map it to an existing\nblock or composition first. If the summary suggests a match, use\n`read_block_type` to inspect the exact schema. Only call something “missing”\nafter checking capabilities, the complete palette, and custom block types.\n\nCommon requirements that are easy to misclassify:\n\n| Requirement | Existing Typeroll primitive |\n|---|---|\n| Full-bleed hero flush below the header | `core/section` + `core/hero`; block-mode sections already own the full width and have no page-shell padding |\n| Responsive icon/card grid | `core/grid` + `core/icon_box`, or `core/feature_grid`; set responsive fields with `set_block_responsive` |\n| Custom cards backed by a collection | `core/repeater` / `core/collection_list` with a site-authored `item_compatible` block type as `item_block` |\n| Grouped collection listing | `core/repeater` with `group_by`; array-valued fields place an item in every matching group |\n| Breadcrumbs in a page template | `template/page_breadcrumbs` (verify that the relevant route supplies a breadcrumb trail) |\n| Generated heading index | `core/table_of_contents`; choose `h2`, `h2-h3`, or `h2-h4` |\n| Previous/next collection item links | `template/item_navigation` in the collection's block template; ordering follows `sort_field` / `sort_dir` |\n| Download CTA that disappears without a file | `template/show_if` around a context-bound `core/button`; a dedicated download block is only editor convenience |\n| Sticky/custom header and multi-column footer | Block-mode header/footer partials plus layout blocks, or one reusable custom block type |\n| Cookie notice | `settings.cookie_consent`, not a page block |\n| Consent copy owned by an installed Extension | `list_extension_installations` → `read_extension_installation` → `update_extension_installation_config`, using exact manifest schema keys |\n| One-off HTML + JavaScript embed | `core/embed`; reusable widgets use a custom block type with `script` |\n| CTA/button variants | `core/cta` and `core/button`, styled from site tokens or a narrowly scoped class |\n| Two-column image/text | `core/media_card`, `core/feature_row`, or `core/columns` |\n| Figures, captions and tables | `core/prose` / richtext; the sanitizer preserves these semantic tags |\n| Full-width block page without a content shell | Native block mode; top-level `core/section` is unconstrained |\n| Brand colors and typography | Site `colors`/`fonts` tokens consumed by core blocks |\n\nReal gaps should say what was checked and why the nearest primitive is not\nenough. “I did not see a dedicated block name” is not evidence by itself.\n\n### 2. Brand + settings\n\nOne `update_site_settings` call with every field you know:\n\n```json\n{\n  \"site_name\": \"Acme Studio\",\n  \"tagline\": \"Short, punchy tagline\",\n  \"language\": \"sv\",\n  \"trailing_slash\": \"always\",\n  \"colors\": {\n    \"primary\":    \"#1a1a2e\",\n    \"secondary\":  \"#16213e\",\n    \"accent\":     \"#e94560\",\n    \"background\": \"#f5f5f5\",\n    \"surface\":    \"#ffffff\",\n    \"text\":       \"#1a1a2e\",\n    \"text_light\": \"#6b7280\"\n  },\n  \"fonts\": { \"heading\": \"Playfair Display\", \"body\": \"Inter\", \"size_base\": 16 },\n  \"contact\": { \"email\": \"hej@acme.se\", \"phone\": \"+46 8 123 456\" },\n  \"social\": { \"instagram\": \"https://instagram.com/acme\" }\n}\n```\n\nRead it back with `read_site_settings`. Google Fonts names are\ncase-sensitive display names (\"Plus Jakarta Sans\", not \"plus jakarta\").\n\n**Site icons are part of brand setup** — upload favicon (32–64px), a\n180×180 apple touch icon, and a 192×192 app icon; set `favicon`,\n`apple_touch_icon`, and `icon_192`. No icon assets? Derive a proposal (see\n`tr-brand`). Also set `settings.logo` to the uploaded brand mark — it feeds\nOG/schema even if the header uses a different lockup.\n\nSet `iframe_allowed_hosts` when customer content embeds a provider outside\nthe built-in YouTube/Vimeo/Google Maps/Calendly set. Values are exact domain\nhostnames, never URLs or wildcards. Read the setting back before deciding that\nan iframe cannot be represented.\n\n### 3. Header + footer partials\n\n**Start from a vetted preset — don't hand-roll the layout.** `read_skill\ntr-header-footer` has robust header + footer presets (centered logo, logo+nav\nwith a no-JS mobile menu, centered + 3-column footers) that avoid the usual\ntraps: clipped logos (no `overflow:hidden` near the logo), distorted logos\n(`height` + `width:auto`), and broken mobile menus. Fill the placeholders and\nrestyle to the palette.\n\nPartials can use either mode. Use HTML for a compact hand-authored nav, or\nblock mode when editors need per-field control or an installed Extension must\nbe placed there. Partials receive the same `{{site.*}}` render context as page\nblocks, so literal brand data is optional rather than required:\n\n```html\n<header class=\"site-header\">\n  <div class=\"header-inner\">\n    <a class=\"header-logo\" href=\"/\"><img src=\"LOGO_MEDIA_URL\" alt=\"Acme Studio\" height=\"40\" /></a>\n    <nav class=\"header-nav\">\n      <a href=\"/om-oss\">Om oss</a>\n      <a href=\"/kontakt\">Kontakt</a>\n    </nav>\n  </div>\n</header>\n<style>\n.site-header{background:var(--color-background);padding:1rem 2rem}\n.header-inner{max-width:1080px;margin:0 auto;display:flex;align-items:center;justify-content:space-between}\n.header-nav{display:flex;gap:2rem}\n.header-nav a{color:var(--color-text);text-decoration:none}\n</style>\n```\n\n`replace_partial partial_id=\"header\" html_content=\"...\"` — same pattern\nfor the footer. Design notes: **no border-bottom on the header if the\nfirst page section should meet it seamlessly** — let background color\nchanges do the separating. Anchor links in nav (`/#section`) are fine.\n\n### 4. Homepage — block tree\n\n`create_page` with the whole tree in one call. Omit block `id`s — the\nplatform assigns them (`blk_…`); you read them back for later\n`update_block` calls.\n\n```\ncreate_page title=\"Start\" slug=\"\" status=\"draft\" content_mode=\"blocks\" blocks=[...]\n```\n\nA proven landing-page skeleton (every top-level block is a `core/section`\n— sections are **natively full-bleed**: the background runs edge-to-edge,\ncontent is constrained by the section's own inner container via the\n`width` field. NEVER use 100vw negative-margin hacks. Anchor ids and\ncustom classes via `style_overrides` are safe on sections from\ntemplate_capabilities_version ≥ 0.15.3; on 0.14.x–0.15.2 they wrap the\nsection in a div and silently break full-bleed — there, put the anchor\non a block *inside* the section instead):\n\n```json\n[\n  { \"type\": \"core/section\", \"data\": { \"background\": \"#ffffff\", \"padding_y\": \"lg\" }, \"children\": [\n    { \"type\": \"core/media_card\", \"data\": {\n        \"image\": \"MEDIA_URL\", \"image_alt\": \"…\", \"image_side\": \"right\",\n        \"heading\": \"Huvudrubriken\", \"heading_level\": \"h2\",\n        \"text\": \"<p>Ingress …</p>\",\n        \"button_label\": \"Kontakta oss\", \"button_url\": \"/#kontakt\"\n    } }\n  ] },\n  { \"type\": \"core/section\", \"data\": { \"padding_y\": \"lg\" }, \"children\": [\n    { \"type\": \"core/heading\", \"data\": { \"text\": \"Så funkar det\", \"level\": \"h2\", \"align\": \"center\" } },\n    { \"type\": \"core/grid\", \"data\": { \"cols\": 3, \"gap\": \"lg\" }, \"children\": [\n      { \"type\": \"core/step_card\", \"data\": { \"number\": \"1\", \"title\": \"…\", \"text\": \"<p>…</p>\" } },\n      { \"type\": \"core/step_card\", \"data\": { \"number\": \"2\", \"title\": \"…\", \"text\": \"<p>…</p>\" } },\n      { \"type\": \"core/step_card\", \"data\": { \"number\": \"3\", \"title\": \"…\", \"text\": \"<p>…</p>\" } }\n    ] }\n  ] },\n  { \"type\": \"core/cta\", \"data\": {\n      \"heading\": \"Redo att börja?\",\n      \"primary_label\": \"Kontakta oss\", \"primary_url\": \"/kontakt\"\n  } }\n]\n```\n\nBlock-palette guidance (verify against `list_block_types` — the source of\ntruth):\n\n- **Hero:** `core/media_card` inside a white section (image beside copy),\n  or `core/hero` (eyebrow/heading/subheading + `primary_*`/`secondary_*`\n  buttons — rendered server-side; `layout: split-right` puts the image\n  beside the text). For a plain text hero: section + heading + prose +\n  button.\n- **Alternating image+text rows (\"zig-zag\" feature/step lists):**\n  `core/feature_row` (template_capabilities_version ≥ 0.29.0) — balanced\n  halves that hug the center gutter so wide screens never open dead air,\n  natural-aspect image, `image_side: left|right` per row, mobile stack\n  order as a field. Prefer it over hand-building `core/columns` with an\n  unbalanced ratio + width-capped text (the classic dead-air trap). Use\n  `core/media_card` instead when you want a *boxed card* with a\n  filled/cropped image.\n- **`core/heading`** decouples `level` (h1–h6, semantics) from `size`\n  (visual) — exactly one `level: h1` per page.\n- **Images:** `core/image` with an uploaded media URL. The build pipeline\n  automatically emits responsive `<picture>` with AVIF/WebP variants\n  (run `generate_image_variants` after upload) — the in-portal preview\n  shows a plain `<img>`, the deployed site gets the upgrade. Use the\n  `radius` field for rounded corners.\n- **Repeaters/listings:** `core/collection_list`, `gallery`,\n  `feature_grid` etc. — alias blocks over `core/repeater`. Use these for\n  collection-driven content instead of hand-writing listing markup. The base\n  repeater also supports `group_by`, group ordering/headings, multi-valued\n  membership, filters, custom `item_block`, and `item_overrides`.\n- **Long-form navigation:** `core/table_of_contents` builds an anchor list\n  from page headings. Collection block templates use\n  `template/item_navigation` for deterministic previous/next links.\n- **Forms:** `create_form`, then place `core/form` with its `form_id`.\n  HTML-mode pages use `<x-form id=\"…\" />`; both paths render the same signed\n  shell and runtime. See `tr-forms`.\n- **`core/html`** is the escape hatch for the genuinely unique thing —\n  not a default. If you reach for it more than once or twice per page,\n  note why (that's block-library feedback).\n\nSlot containers (`core/columns`, `core/tabs`): populate them either by\npassing the whole tree inline (`block={ type: 'core/columns', slots:\n[[…],[…]] }`) or incrementally with `add_block parent_id=<columns-id>\nslot_index=0|1`. Requires template_capabilities_version ≥ 0.15.2 — on\nolder sites use `core/grid` (children flow into columns) instead.\n\nPartial last rows (template_capabilities_version ≥ 0.16.5): when N equal\ncards don't divide by the grid's column count (5 cards, 3 cols), set\n`last_row: 'center'` on the `core/grid` — the orphan row auto-centers.\nNEVER invent a \"wide\" variant of one peer card to fill the hole, and\ndon't hand-roll 6-column CSS tricks; both distort content to patch\nlayout. On older sites, pick a column count that divides N.\n\nIcons (template_capabilities_version ≥ 0.16.0): `type: 'icon'` fields on\n`core/icon`, `core/icon_box`, and `core/step_card` render inline SVG when\nthe value is a name from `get_site_capabilities → core_icon_names` (a\ncurated Lucide subset — `check`, `star`, `shield-check`, `mail`,\n`arrow-right`, `truck`, `chart-line`, …). Any other value (emoji, plain\ntext) renders as text, so emoji stand-ins keep working. Icons inherit\nsize from font-size and color from `currentColor`/the block's color\nfield. On older sites icons don't render — use emoji or CSS markers.\n\nEditor schemas (template_capabilities_version ≥ 0.39.0): labelled enum\noptions, newline-edited string lists, nested object/repeating-array fields,\nand internal-page pickers for URL fields are first-class. Do not flatten an\nExtension's `props_schema` merely to make it editable.\n\nFor an Extension placed in an HTML header or footer, require\n`supports_extension_html_partial_directive: true`. The broader\n`supports_extension_html_directive` flag covers HTML page bodies and is not\nproof that an older static build expands partial directives.\n\nFor a cross-page Extension flow, require both\n`supports_extension_site_navigation` and `supports_extension_storage`. Use\n`context.site.navigate(\"/path/\")` and installation-scoped\n`context.storage.session` rather than direct root-path navigation, Web Storage,\nor personal data in query parameters. This keeps navigation inside the current\npreview and preserves state without exposing it through URLs or referrers.\n\nKnown limitations (honest list — don't fight them):\n\n- **`core/tabs` label icons don't render** (the tab strip is built\n  client-side without the icon pipeline). Text labels only.\n\nTheming: block primitives render neutral. Brand color/typography comes\nfrom settings (step 2). For page-specific polish (e.g. a colored card\ntreatment), a single `core/html` block with a small `<style>` scoped to\n`[data-bid]`/section selectors is acceptable — keep it minimal and note\nit in your log.\n\n### 5. Inner pages\n\nSame pattern: `create_page` with `content_mode: \"blocks\"` and a section\ntree. Standard set: Om oss, Tjänster, Kontakt — or what the brief says.\nDefault new pages to `status: \"draft\"`; publish after review.\n\n### 5b. If the legacy site is still live, scrape canonical content\n\nFor pages with canonical text (privacy policy, terms, about), `WebFetch`\nthe live page and carry the text verbatim — don't rewrite legal copy\nfrom memory. Convert to prose blocks (or one `core/html` for complex\nlegacy markup).\n\n### 6. Preview + iterate\n\n```\nget_preview_link                  # signed URL for browser review\nget_page_preview page_id=\"home\"   # rendered HTML for structural checks\n```\n\n**Self-review the visuals before you call it done — appearance AND\nreadability, not just structure.** Screenshot the deployed/preview site at\ndesktop (~1440px) and mobile (~390px) and look: logo FULLY VISIBLE (not clipped by\na header's overflow:hidden) + legible + brand-compliant against its actual\nbackground — screenshot the header IN CONTEXT, not the logo element in isolation\n(an element shot hides layout clipping); a light wordmark must not sit bare on a\nlight surface. Text contrast everywhere, no horizontal scroll or mid-word\nbreaks, every image rendered, mobile layout actually collapsed. \"No overflow +\ncopy present\" is not a design review — never report a build as done/perfect off\nstructural metrics alone.\n\nShare the preview link. Iterate on feedback with `update_block` /\n`add_block` / `move_block` — that's the point of block mode: surgical\nedits, not full-page rewrites.\n\n### 7. Deploy\n\nWhen the user approves:\n```\ntrigger_deploy\nget_deploy_status job_id=<id>\n```\n\n## Secondary path: HTML mode\n\nFor migrated legacy pages or a hand-crafted one-off, `content_mode:\n\"html\"` still works. Rules that apply there (and only there):\n\n- Wrap the body in a page-scope `<article class=\"my-page\">` and prefix\n  selectors with it — the `.page-content` shell sets width/padding at\n  normal specificity.\n- HTML-mode bodies are container-constrained; full-bleed requires the\n  negative-margin escape (`margin-left: calc(50% - 50vw); width: 100vw`)\n  — never combine with `overflow-x: clip` on a wrapper.\n- `set_page_mode` flips a page between modes;\n  `convert_page_to_blocks` does a heuristic HTML→blocks conversion.\n\n## Pitfalls\n\n- **Don't create pages that already exist** — `list_pages` first; use\n  `update_page` if the slug is taken.\n- **`update_page` takes a `patch` object**, not flat fields.\n- **Don't hardcode block field names from memory** — schemas are\n  per-site (`list_block_types`/`read_block_type`).\n- **One `level: h1` per page** (core/heading) — SEO + screen readers.\n- **Don't simplify data during import** — preserve Swedish characters in\n  labels (`affärsutveckling`, not the ASCII-folded slug), keep titles\n  verbatim; slugs are derived for URLs only.\n- **Minimal JS.** Inline `<script>` in page content is stripped by the\n  sanitizer. One-off interactivity uses `core/embed`; reusable interactivity\n  uses a block-type `script`; site-wide code belongs in `scripts_body_end`.\n",
+    "tr-new-site": "---\nname: tr-new-site\ndescription: Use when the user wants to create a new Typeroll site from scratch, set up the initial design, or bootstrap a blank site with working header/footer, brand colors, and a homepage. Also triggers on \"start a new site\", \"set up a site for\", or \"build a website for [company]\".\n---\n\n# Bootstrap a new Typeroll site\n\n> **The buffer model (draft writes).** Every content write in this recipe\n> (pages, blocks, partials, collection items) lands in an unsaved per-doc\n> DRAFT — deploys and plain previews only see SAVED content. For recipe-style\n> build work, pass `save: true` on write calls (the work is pre-approved by\n> the task itself), or run `commit_working_copy` per doc before any\n> `trigger_deploy`. Preview your drafts with `include_working_copy: true`.\n\n\nStart here when the site already exists as a database record (created via\nthe portal UI or API) but has no design, no header/footer, and no pages.\nThe goal is to go from blank to a working 4-page site with correct brand\nidentity in a single session.\n\n**Pages are built in block mode** — the platform default. Blocks give\nstructured, per-field editing, native full-bleed sections, responsive\nbreakpoints, and templates. HTML mode is the secondary path for\nhand-crafted one-offs and migrated content (section at the end).\n\n## Preconditions\n\n- `@typeroll/mcp-server` configured with a valid `TYPEROLL_API_KEY`.\n- The site exists (confirm with `get_site`).\n- You have the customer brief: company name, industry, 2–3 key brand colors,\n  tone of voice, and a list of initial pages.\n\n## Recipe\n\n### 1. Audit current state\n\n```\nget_site\nget_site_capabilities   # template_capabilities_version — what this deployment supports\nread_site_settings      # see what (if anything) is already configured\nlist_pages              # don't overwrite pages that already exist\nlist_partials           # check if header/footer already have content\nlist_block_types        # the per-site block palette — NEVER assume, always list\n```\n\n**Do not choose a content mode or file a platform gap before these calls\nreturn.** For every required visual/behavioral element, map it to an existing\nblock or composition first. If the summary suggests a match, use\n`read_block_type` to inspect the exact schema. Only call something “missing”\nafter checking capabilities, the complete palette, and custom block types.\n\nCommon requirements that are easy to misclassify:\n\n| Requirement | Existing Typeroll primitive |\n|---|---|\n| Full-bleed hero flush below the header | `core/section` + `core/hero`; block-mode sections already own the full width and have no page-shell padding |\n| Responsive icon/card grid | `core/grid` + `core/icon_box`, or `core/feature_grid`; set responsive fields with `set_block_responsive` |\n| Custom cards backed by a collection | `core/repeater` / `core/collection_list` with a site-authored `item_compatible` block type as `item_block` |\n| Grouped collection listing | `core/repeater` with `group_by`; array-valued fields place an item in every matching group |\n| Breadcrumbs in a page template | `template/page_breadcrumbs` (verify that the relevant route supplies a breadcrumb trail) |\n| Generated heading index | `core/table_of_contents`; choose `h2`, `h2-h3`, or `h2-h4` |\n| Previous/next collection item links | `template/item_navigation` in the collection's block template; ordering follows `sort_field` / `sort_dir` |\n| Download CTA that disappears without a file | `template/show_if` around a context-bound `core/button`; a dedicated download block is only editor convenience |\n| Sticky/custom header and multi-column footer | Block-mode header/footer partials plus layout blocks, or one reusable custom block type |\n| Cookie notice | `settings.cookie_consent`, not a page block |\n| Consent copy owned by an installed Extension | `list_extension_installations` → `read_extension_installation` → `update_extension_installation_config`, using exact manifest schema keys; the update queues a production deploy by default |\n| One-off HTML + JavaScript embed | `core/embed`; reusable widgets use a custom block type with `script` |\n| CTA/button variants | `core/cta` and `core/button`, styled from site tokens or a narrowly scoped class |\n| Two-column image/text | `core/media_card`, `core/feature_row`, or `core/columns` |\n| Figures, captions and tables | `core/prose` / richtext; the sanitizer preserves these semantic tags |\n| Full-width block page without a content shell | Native block mode; top-level `core/section` is unconstrained |\n| Brand colors and typography | Site `colors`/`fonts` tokens consumed by core blocks |\n\nReal gaps should say what was checked and why the nearest primitive is not\nenough. “I did not see a dedicated block name” is not evidence by itself.\n\n### 2. Brand + settings\n\nOne `update_site_settings` call with every field you know:\n\n```json\n{\n  \"site_name\": \"Acme Studio\",\n  \"tagline\": \"Short, punchy tagline\",\n  \"language\": \"sv\",\n  \"trailing_slash\": \"always\",\n  \"colors\": {\n    \"primary\":    \"#1a1a2e\",\n    \"secondary\":  \"#16213e\",\n    \"accent\":     \"#e94560\",\n    \"background\": \"#f5f5f5\",\n    \"surface\":    \"#ffffff\",\n    \"text\":       \"#1a1a2e\",\n    \"text_light\": \"#6b7280\"\n  },\n  \"fonts\": { \"heading\": \"Playfair Display\", \"body\": \"Inter\", \"size_base\": 16 },\n  \"contact\": { \"email\": \"hej@acme.se\", \"phone\": \"+46 8 123 456\" },\n  \"social\": { \"instagram\": \"https://instagram.com/acme\" }\n}\n```\n\nRead it back with `read_site_settings`. Google Fonts names are\ncase-sensitive display names (\"Plus Jakarta Sans\", not \"plus jakarta\").\n\n**Site icons are part of brand setup** — upload favicon (32–64px), a\n180×180 apple touch icon, and a 192×192 app icon; set `favicon`,\n`apple_touch_icon`, and `icon_192`. No icon assets? Derive a proposal (see\n`tr-brand`). Also set `settings.logo` to the uploaded brand mark — it feeds\nOG/schema even if the header uses a different lockup.\n\nSet `iframe_allowed_hosts` when customer content embeds a provider outside\nthe built-in YouTube/Vimeo/Google Maps/Calendly set. Values are exact domain\nhostnames, never URLs or wildcards. Read the setting back before deciding that\nan iframe cannot be represented.\n\n### 3. Header + footer partials\n\n**Start from a vetted preset — don't hand-roll the layout.** `read_skill\ntr-header-footer` has robust header + footer presets (centered logo, logo+nav\nwith a no-JS mobile menu, centered + 3-column footers) that avoid the usual\ntraps: clipped logos (no `overflow:hidden` near the logo), distorted logos\n(`height` + `width:auto`), and broken mobile menus. Fill the placeholders and\nrestyle to the palette.\n\nPartials can use either mode. Use HTML for a compact hand-authored nav, or\nblock mode when editors need per-field control or an installed Extension must\nbe placed there. Partials receive the same `{{site.*}}` render context as page\nblocks, so literal brand data is optional rather than required:\n\n```html\n<header class=\"site-header\">\n  <div class=\"header-inner\">\n    <a class=\"header-logo\" href=\"/\"><img src=\"LOGO_MEDIA_URL\" alt=\"Acme Studio\" height=\"40\" /></a>\n    <nav class=\"header-nav\">\n      <a href=\"/om-oss\">Om oss</a>\n      <a href=\"/kontakt\">Kontakt</a>\n    </nav>\n  </div>\n</header>\n<style>\n.site-header{background:var(--color-background);padding:1rem 2rem}\n.header-inner{max-width:1080px;margin:0 auto;display:flex;align-items:center;justify-content:space-between}\n.header-nav{display:flex;gap:2rem}\n.header-nav a{color:var(--color-text);text-decoration:none}\n</style>\n```\n\n`replace_partial partial_id=\"header\" html_content=\"...\"` — same pattern\nfor the footer. Design notes: **no border-bottom on the header if the\nfirst page section should meet it seamlessly** — let background color\nchanges do the separating. Anchor links in nav (`/#section`) are fine.\n\n### 4. Homepage — block tree\n\n`create_page` with the whole tree in one call. Omit block `id`s — the\nplatform assigns them (`blk_…`); you read them back for later\n`update_block` calls.\n\n```\ncreate_page title=\"Start\" slug=\"\" status=\"draft\" content_mode=\"blocks\" blocks=[...]\n```\n\nA proven landing-page skeleton (every top-level block is a `core/section`\n— sections are **natively full-bleed**: the background runs edge-to-edge,\ncontent is constrained by the section's own inner container via the\n`width` field. NEVER use 100vw negative-margin hacks. Anchor ids and\ncustom classes via `style_overrides` are safe on sections from\ntemplate_capabilities_version ≥ 0.15.3; on 0.14.x–0.15.2 they wrap the\nsection in a div and silently break full-bleed — there, put the anchor\non a block *inside* the section instead):\n\n```json\n[\n  { \"type\": \"core/section\", \"data\": { \"background\": \"#ffffff\", \"padding_y\": \"lg\" }, \"children\": [\n    { \"type\": \"core/media_card\", \"data\": {\n        \"image\": \"MEDIA_URL\", \"image_alt\": \"…\", \"image_side\": \"right\",\n        \"heading\": \"Huvudrubriken\", \"heading_level\": \"h2\",\n        \"text\": \"<p>Ingress …</p>\",\n        \"button_label\": \"Kontakta oss\", \"button_url\": \"/#kontakt\"\n    } }\n  ] },\n  { \"type\": \"core/section\", \"data\": { \"padding_y\": \"lg\" }, \"children\": [\n    { \"type\": \"core/heading\", \"data\": { \"text\": \"Så funkar det\", \"level\": \"h2\", \"align\": \"center\" } },\n    { \"type\": \"core/grid\", \"data\": { \"cols\": 3, \"gap\": \"lg\" }, \"children\": [\n      { \"type\": \"core/step_card\", \"data\": { \"number\": \"1\", \"title\": \"…\", \"text\": \"<p>…</p>\" } },\n      { \"type\": \"core/step_card\", \"data\": { \"number\": \"2\", \"title\": \"…\", \"text\": \"<p>…</p>\" } },\n      { \"type\": \"core/step_card\", \"data\": { \"number\": \"3\", \"title\": \"…\", \"text\": \"<p>…</p>\" } }\n    ] }\n  ] },\n  { \"type\": \"core/cta\", \"data\": {\n      \"heading\": \"Redo att börja?\",\n      \"primary_label\": \"Kontakta oss\", \"primary_url\": \"/kontakt\"\n  } }\n]\n```\n\nBlock-palette guidance (verify against `list_block_types` — the source of\ntruth):\n\n- **Hero:** `core/media_card` inside a white section (image beside copy),\n  or `core/hero` (eyebrow/heading/subheading + `primary_*`/`secondary_*`\n  buttons — rendered server-side; `layout: split-right` puts the image\n  beside the text). For a plain text hero: section + heading + prose +\n  button.\n- **Alternating image+text rows (\"zig-zag\" feature/step lists):**\n  `core/feature_row` (template_capabilities_version ≥ 0.29.0) — balanced\n  halves that hug the center gutter so wide screens never open dead air,\n  natural-aspect image, `image_side: left|right` per row, mobile stack\n  order as a field. Prefer it over hand-building `core/columns` with an\n  unbalanced ratio + width-capped text (the classic dead-air trap). Use\n  `core/media_card` instead when you want a *boxed card* with a\n  filled/cropped image.\n- **`core/heading`** decouples `level` (h1–h6, semantics) from `size`\n  (visual) — exactly one `level: h1` per page.\n- **Images:** `core/image` with an uploaded media URL. The build pipeline\n  automatically emits responsive `<picture>` with AVIF/WebP variants\n  (run `generate_image_variants` after upload) — the in-portal preview\n  shows a plain `<img>`, the deployed site gets the upgrade. Use the\n  `radius` field for rounded corners.\n- **Repeaters/listings:** `core/collection_list`, `gallery`,\n  `feature_grid` etc. — alias blocks over `core/repeater`. Use these for\n  collection-driven content instead of hand-writing listing markup. The base\n  repeater also supports `group_by`, group ordering/headings, multi-valued\n  membership, filters, custom `item_block`, and `item_overrides`.\n- **Long-form navigation:** `core/table_of_contents` builds an anchor list\n  from page headings. Collection block templates use\n  `template/item_navigation` for deterministic previous/next links.\n- **Forms:** `create_form`, then place `core/form` with its `form_id`.\n  HTML-mode pages use `<x-form id=\"…\" />`; both paths render the same signed\n  shell and runtime. See `tr-forms`.\n- **`core/html`** is the escape hatch for the genuinely unique thing —\n  not a default. If you reach for it more than once or twice per page,\n  note why (that's block-library feedback).\n\nSlot containers (`core/columns`, `core/tabs`): populate them either by\npassing the whole tree inline (`block={ type: 'core/columns', slots:\n[[…],[…]] }`) or incrementally with `add_block parent_id=<columns-id>\nslot_index=0|1`. Requires template_capabilities_version ≥ 0.15.2 — on\nolder sites use `core/grid` (children flow into columns) instead.\n\nPartial last rows (template_capabilities_version ≥ 0.16.5): when N equal\ncards don't divide by the grid's column count (5 cards, 3 cols), set\n`last_row: 'center'` on the `core/grid` — the orphan row auto-centers.\nNEVER invent a \"wide\" variant of one peer card to fill the hole, and\ndon't hand-roll 6-column CSS tricks; both distort content to patch\nlayout. On older sites, pick a column count that divides N.\n\nIcons (template_capabilities_version ≥ 0.16.0): `type: 'icon'` fields on\n`core/icon`, `core/icon_box`, and `core/step_card` render inline SVG when\nthe value is a name from `get_site_capabilities → core_icon_names` (a\ncurated Lucide subset — `check`, `star`, `shield-check`, `mail`,\n`arrow-right`, `truck`, `chart-line`, …). Any other value (emoji, plain\ntext) renders as text, so emoji stand-ins keep working. Icons inherit\nsize from font-size and color from `currentColor`/the block's color\nfield. On older sites icons don't render — use emoji or CSS markers.\n\nEditor schemas (template_capabilities_version ≥ 0.39.0): labelled enum\noptions, newline-edited string lists, nested object/repeating-array fields,\nand internal-page pickers for URL fields are first-class. Do not flatten an\nExtension's `props_schema` merely to make it editable.\n\nFor an Extension placed in an HTML header or footer, require\n`supports_extension_html_partial_directive: true`. The broader\n`supports_extension_html_directive` flag covers HTML page bodies and is not\nproof that an older static build expands partial directives.\n\nFor a cross-page Extension flow, require both\n`supports_extension_site_navigation` and `supports_extension_storage`. Use\n`context.site.navigate(\"/path/\")` and installation-scoped\n`context.storage.session` rather than direct root-path navigation, Web Storage,\nor personal data in query parameters. This keeps navigation inside the current\npreview and preserves state without exposing it through URLs or referrers.\n\nKnown limitations (honest list — don't fight them):\n\n- **`core/tabs` label icons don't render** (the tab strip is built\n  client-side without the icon pipeline). Text labels only.\n\nTheming: block primitives render neutral. Brand color/typography comes\nfrom settings (step 2). For page-specific polish (e.g. a colored card\ntreatment), a single `core/html` block with a small `<style>` scoped to\n`[data-bid]`/section selectors is acceptable — keep it minimal and note\nit in your log.\n\n### 5. Inner pages\n\nSame pattern: `create_page` with `content_mode: \"blocks\"` and a section\ntree. Standard set: Om oss, Tjänster, Kontakt — or what the brief says.\nDefault new pages to `status: \"draft\"`; publish after review.\n\n### 5b. If the legacy site is still live, scrape canonical content\n\nFor pages with canonical text (privacy policy, terms, about), `WebFetch`\nthe live page and carry the text verbatim — don't rewrite legal copy\nfrom memory. Convert to prose blocks (or one `core/html` for complex\nlegacy markup).\n\n### 6. Preview + iterate\n\n```\nget_preview_link                  # signed URL for browser review\nget_page_preview page_id=\"home\"   # rendered HTML for structural checks\n```\n\n**Self-review the visuals before you call it done — appearance AND\nreadability, not just structure.** Screenshot the deployed/preview site at\ndesktop (~1440px) and mobile (~390px) and look: logo FULLY VISIBLE (not clipped by\na header's overflow:hidden) + legible + brand-compliant against its actual\nbackground — screenshot the header IN CONTEXT, not the logo element in isolation\n(an element shot hides layout clipping); a light wordmark must not sit bare on a\nlight surface. Text contrast everywhere, no horizontal scroll or mid-word\nbreaks, every image rendered, mobile layout actually collapsed. \"No overflow +\ncopy present\" is not a design review — never report a build as done/perfect off\nstructural metrics alone.\n\nShare the preview link. Iterate on feedback with `update_block` /\n`add_block` / `move_block` — that's the point of block mode: surgical\nedits, not full-page rewrites.\n\n### 7. Deploy\n\nWhen the user approves:\n```\ntrigger_deploy\nget_deploy_status job_id=<id>\n```\n\n## Secondary path: HTML mode\n\nFor migrated legacy pages or a hand-crafted one-off, `content_mode:\n\"html\"` still works. Rules that apply there (and only there):\n\n- Wrap the body in a page-scope `<article class=\"my-page\">` and prefix\n  selectors with it — the `.page-content` shell sets width/padding at\n  normal specificity.\n- HTML-mode bodies are container-constrained; full-bleed requires the\n  negative-margin escape (`margin-left: calc(50% - 50vw); width: 100vw`)\n  — never combine with `overflow-x: clip` on a wrapper.\n- `set_page_mode` flips a page between modes;\n  `convert_page_to_blocks` does a heuristic HTML→blocks conversion.\n\n## Pitfalls\n\n- **Don't create pages that already exist** — `list_pages` first; use\n  `update_page` if the slug is taken.\n- **`update_page` takes a `patch` object**, not flat fields.\n- **Don't hardcode block field names from memory** — schemas are\n  per-site (`list_block_types`/`read_block_type`).\n- **One `level: h1` per page** (core/heading) — SEO + screen readers.\n- **Don't simplify data during import** — preserve Swedish characters in\n  labels (`affärsutveckling`, not the ASCII-folded slug), keep titles\n  verbatim; slugs are derived for URLs only.\n- **Minimal JS.** Inline `<script>` in page content is stripped by the\n  sanitizer. One-off interactivity uses `core/embed`; reusable interactivity\n  uses a block-type `script`; site-wide code belongs in `scripts_body_end`.\n",
     "tr-page-template": "---\nname: tr-page-template\ndescription: Use when several pages on a Typeroll site share the same outer structure — category pages, service-detail pages, landing-page variants — and the user wants to edit the shared bits in one place. Covers two flows: the HTML-mode pattern with partials + `<x-include>` (works everywhere, recommended for Phase 1), and the formal block-mode PageTemplate (when the site is in block mode).\n---\n\n# Share structure across pages\n\nWhen you have N pages that follow the same skeleton — say 7 category landing pages, each with `Hero → Intro → Features grid → CTA banner` — you don't want to edit 7 HTML bodies whenever the design changes. There are two ways to share structure in Typeroll, and the right pick depends on the site's content mode.\n\n## Decide which pattern fits\n\n| Site is mostly in… | Use |\n|---|---|\n| **HTML mode** (the Phase 1 default — most sites) | Partials + `<x-include>` — pattern A below |\n| **Block mode** (`supports_blocks_mode=true` and the page's `content_mode='blocks'`) | PageTemplate via `set_page_template` — pattern B below |\n\nYou can check the site's mode by reading any existing page — `content_mode` is a top-level field on the page doc.\n\nIf unsure, default to **pattern A**. It works on every Typeroll site and the migration to block-mode templates later is mechanical.\n\n## Pattern A — Partials + `<x-include>` (HTML mode)\n\nA partial is a named HTML fragment. Putting `<x-include name=\"my-partial\" />` anywhere in a page body inlines that fragment at build time. Editing the partial updates every page that references it on the next deploy.\n\n### Recipe for \"7 category pages with shared structure\"\n\n#### 1. Identify the shared chunks\n\nWalk through one category page and mark which sections are the same across all 7:\n\n```\n[Hero: title + tagline + image]            ← varies per category\n[Intro paragraph]                          ← varies per category\n[Common: \"Why choose us\" three-tile band]  ← identical across 7\n[Common: CTA banner with newsletter form]  ← identical across 7\n[Common: footer testimonials]              ← identical across 7\n```\n\nThree reusable bits: `why-choose-us`, `cta-newsletter`, `footer-testimonials`.\n\n#### 2. Create the partials\n\n```\ncreate_partial partial_id=\"why-choose-us\" html_content=\"<section class=\\\"why\\\">\n  <div class=\\\"container\\\">\n    <h2>Varför oss</h2>\n    <div class=\\\"why__grid\\\">\n      <div class=\\\"why__tile\\\"><h3>Erfarenhet</h3><p>20 år i branschen.</p></div>\n      <div class=\\\"why__tile\\\"><h3>Kvalitet</h3><p>Vi mäter på allt.</p></div>\n      <div class=\\\"why__tile\\\"><h3>Närhet</h3><p>Lokala kontor i tre städer.</p></div>\n    </div>\n  </div>\n</section>\n<style>\n.why{padding:4rem 0;background:var(--color-surface)}\n.why__grid{display:grid;grid-template-columns:repeat(auto-fit,minmax(220px,1fr));gap:2rem;margin-top:2rem}\n.why__tile h3{font-family:var(--font-heading);margin-bottom:0.5rem}\n</style>\"\n```\n\nSame for `cta-newsletter` and `footer-testimonials`.\n\nThe partial-id is what `<x-include>` references — keep it kebab-case and descriptive.\n\n#### 3. Build each category page using the partials\n\n```\ncreate_page title=\"Tjänster för bostadsrätter\" slug=\"brf\" content_mode=\"html\"\n  html_content=\"<section class=\\\"hero hero--brf\\\">\n  <div class=\\\"container\\\">\n    <h1>Tjänster för bostadsrättsföreningar</h1>\n    <p class=\\\"hero__tagline\\\">Trygg förvaltning, helt utan överraskningar.</p>\n  </div>\n</section>\n\n<section class=\\\"intro\\\">\n  <div class=\\\"container\\\">\n    <p>Vi har förvaltat över 200 BRF:er i Stockholmsområdet sedan 2005.</p>\n  </div>\n</section>\n\n<x-include name=\\\"why-choose-us\\\" />\n<x-include name=\\\"cta-newsletter\\\" />\n<x-include name=\\\"footer-testimonials\\\" />\"\n```\n\nRepeat for the other 6 categories, varying only the `hero` + `intro` sections.\n\n#### 4. Edit once, propagate everywhere\n\nAdding a fourth \"Why us\" tile? Edit `why-choose-us` once:\n\n```\nreplace_partial partial_id=\"why-choose-us\" html_content=\"<section class=\\\"why\\\">\n  ...four tiles instead of three...\n</section>\"\n```\n\nAll 7 pages pick up the change on the next deploy. No multi-page diff.\n\n### Edge cases\n\n- **`<x-include>` self-closes or has an explicit close.** Both forms work: `<x-include name=\"x\" />` and `<x-include name=\"x\"></x-include>`.\n- **Nested includes are NOT supported.** If `partial-a` references `<x-include name=\"partial-b\" />`, the inner reference is not expanded. Flatten the hierarchy at design time.\n- **Unknown partial → silently empty.** A typo in the `name` attribute removes the tag at expand time with no warning. Verify the partial id matches one returned by `list_partials`.\n- **Partial content is sanitised on save.** `<script>` tags get stripped at partial creation, then the inlined HTML is sanitised again when the page renders. Two passes; both intentional.\n\n## Pattern B — Formal PageTemplate (block mode)\n\nAvailable when the site is in block mode (`supports_blocks_mode=true` and the page's `content_mode='blocks'`). A PageTemplate is a `Block[]` tree with one or more `template_content_slot` blocks marking where the page's own blocks go.\n\n### Recipe\n\n#### 1. List existing templates\n\n```\nlist_page_templates\n```\n\nReturns templates already defined on the site.\n\n#### 2. Create / pick a template\n\nTemplates live as `Block[]` trees under `paths.pageTemplate(orgId, siteId, templateId)`. They're created via the portal UI for now (no dedicated MCP `create_page_template` tool yet — the chat-tool surface in `anthropic.ts` exposes list + set, not create).\n\nIf the template you want doesn't exist, ask the user to create it via **/app/sites/{id}/templates** in the portal, or use the block-mode-aware AI chat in the portal which knows how to write template trees.\n\n#### 3. Assign it to a page\n\n```\nset_page_template page_id=\"brf\" template_id=\"category-landing\"\n```\n\nThe renderer composes the template's blocks with the page's blocks at build time via `composePageWithTemplate` (replacing each `template_content_slot` with the page's own block tree). The page's CSS / theme / SEO settings are unchanged.\n\n#### 4. Editing the template propagates to every page\n\nThe same template id can be set on multiple pages. Editing the template updates them all on next deploy.\n\n### When B beats A\n\n- Block-mode editor surfaces template assignment as a dropdown in the page settings.\n- Templates compose with the block tree, so the page author can drop content blocks into named slots rather than writing HTML.\n- The renderer is aware of block CSS / JS bundling — assets per block type are aggregated automatically.\n\n### When A still beats B even in block mode\n\n- Need a small reusable HTML chunk in the middle of a block-mode page → still use `<x-include>` (block-mode bodies can contain free HTML blocks that include partials).\n- One-page change wanted without affecting siblings → just edit the page's blocks; templates are all-or-nothing.\n\n## Refactor: 7 already-existing pages into a shared partial\n\nIf the 7 pages already exist as separate HTML bodies with duplicated chunks:\n\n1. `read_page page_id=<one of them>` and identify the shared HTML literally — character-for-character chunks that repeat across all 7.\n2. `create_partial partial_id=<descriptive-name> html_content=\"<the shared chunk>\"`.\n3. For each of the 7 pages, `update_page` with the shared chunk replaced by `<x-include name=\"<name>\" />`.\n4. Deploy. The rendered output should be byte-identical to before; only the source pages got shorter.\n\nDon't try to abstract first time and hand-write the partials. Refactor from real, working duplication.\n\n## Pitfalls\n\n- **Don't put hero / page-specific content into a shared partial.** A partial is for things that are *truly identical* across uses. The moment you want it to vary by page, lift the differing parts back into the page body.\n- **Header and footer already use partials.** Don't recreate them inside a category-page shared partial — they're injected by the layout, not by the page body.\n- **CSS scoping.** Partials carrying `<style>` blocks merge into every page that includes them. If two partials define `.tile` differently, the last one wins. Either namespace classes per partial (`.why__tile`, `.feature__tile`) or move shared styles into the site's global CSS via `update_site_settings` → `custom_css`.\n- **Don't migrate to block-mode templates just because you can.** Phase 1 sites are HTML-mode by design; partials + includes give you 90% of the value with zero risk.\n- **`<x-include>` in a partial body referencing another partial.** Won't expand (see edge cases). If you find yourself wanting this, you're building a layout system inside the partial system — at that point the site probably wants block mode.\n",
     "tr-redesign-branch": "---\nname: tr-redesign-branch\ndescription: Use when the user asks to redesign, modernize, or restructure a Typeroll site (or a section of it). Forces branch-isolated work so the live site stays untouched until the redesign is approved.\n---\n\n# Redesign a site without breaking the live one\n\n> **The buffer model (draft writes).** Every content write in this recipe\n> (pages, blocks, partials, collection items) lands in an unsaved per-doc\n> DRAFT — deploys and plain previews only see SAVED content. For recipe-style\n> build work, pass `save: true` on write calls (the work is pre-approved by\n> the task itself), or run `commit_working_copy` per doc before any\n> `trigger_deploy`. Preview your drafts with `include_working_copy: true`.\n\n\nSite-wide changes are exactly where copy-on-write branches earn their\nkeep. This skill enforces the discipline: every redesign happens on a\nbranch, preview-checked end-to-end, merged only after user sign-off.\n\n**Copy comes from the LIVE page, not a local draft.** A redesign changes\nthe design, not the words. Read the existing copy from the live page\n(`read_page`/`batch_read_pages`) and carry it over verbatim. Local\n`sources/*.md` files are drafts — use them only if the user explicitly\nsays \"apply the copy in `<file>`\". Don't invent new headlines, drop\nsections, or \"restore\" text from an old draft; a draft that had drifted\nfrom the live page once sent a whole redesign off the approved wording.\nWhen you must change a word, change it on the live page too and keep the\ndraft file in sync.\n\n**Restraint beats decoration.** Default to clean, purposeful design. Don't reach\nfor decorative motifs (suns, blobs, glows, mascots, confetti) to look \"graphic\" —\nunless a motif *means something for this brand/page*, it reads as random, and it's\nusually the exact thing that clips, seams, and crops. These fragile patterns broke\na real build — avoid them:\n- **A shape divider (wave/curve) between two sections** → don't hand-roll it in\n  `core/html`; a separate stacked shape seams against the next section (a Chrome\n  sub-pixel hairline). Use `core/section`'s **`divider_top` / `divider_bottom`**\n  (`wave | curve | tilt`) — the platform paints it in the section's own colour and\n  overlaps the neighbour by 1px, so it's seam-free by construction. Put the divider\n  on the section whose colour should rise/dip into the neighbour.\n- **A glow/decoration inside an `overflow:hidden` box** → clipped to a hard edge.\n  Put it in a non-clipped layer, or size it to fade out before the box edge.\n- **`object-fit:cover` on a portrait inside a circle/frame** → crops heads and\n  faces. Use `contain`, reframe the source art, or size the frame to the art.\n- **A gradient \"fade\" at a section join** → reads as the design being cut off.\n  Make transitions deliberate (a clean shape or a solid edge), never a fade.\nThese are invisible in a small full-page thumbnail and only show at real size in\nthe actual browser — see the review gate in step 6.\n\n## Recipe\n\n### 1. Discover (always)\n\n```\nget_site\nread_site_settings\nread_partial partial_id=\"header\"\nread_partial partial_id=\"footer\"\nlist_pages limit=20\nbatch_read_pages page_ids=[<top 3-5 pages>]   # see actual conventions\nlist_partials                                  # what free blocks exist\nlist_collections                               # any data we need to consider\n```\n\nWrite the user a short read-back: *\"This is a 12-page agency site\nusing CSS variables, primary color #1e40af, Inter heading + Source\nSans body. Existing pages are content-dense, single-column. Main nav\nhas 5 items including a CTA. I'd suggest...\"*\n\nConfirm direction before touching anything.\n\n### 2. Create a branch\n\n```\ncreate_branch name=\"<descriptive name>\"\n```\n\nSave the response's `id` — pass it as `version=<id>` on every\nsubsequent call. Branches default `robots_blocked: true` so a\nhalf-finished redesign won't be indexed.\n\n### 2b. Write the design spec (REQUIRED — before you build)\n\nEvery redesign branch MUST carry a written design spec. Without it the\ndesign choices live only in the agent's head, so a later \"just tweak the\nillustration style / palette\" means re-deriving everything by hand (this\ngap cost a real project a full reverse-engineering pass). Write it BEFORE\nbuilding so it guides the work, and keep it in sync as the design evolves.\n\nSave it as a markdown doc with the project (e.g. `design-spec.md`, or\n`prompts/design-system.md`) — or, if there's no local working dir, as an\nunlisted page on the branch. It must capture:\n\n- **Palette** — every role + hex (background, surface, primary, accent,\n  text, borders), and where the variant *diverges* from the brand and why.\n- **Typography** — fonts + weights/sizes per role.\n- **Illustration / imagery style** — the exact image-gen prompt prefix\n  (tone, palette, formspråk, framing rules), so the imagery can be\n  regenerated or restyled on its own without touching layout. State the\n  business/concept constraints the imagery must respect (a wrong-concept\n  image is worse than none).\n- **Section structure** — the page's sections in order + each one's\n  treatment (band colour, layout).\n- **Rationale** — one line per major choice: *why* this direction.\n\nWhen the user later says \"adjust just the illustrations\" or \"change the\npalette\", you edit the spec first, then apply — the spec is the source of\ntruth for the design intent.\n\n### 3. Iterate on the branch\n\nFor each redesign step:\n\na. Make the change with `?version=<branch-id>`. Updates here don't\n   touch main:\n\n   ```\n   update_partial partial_id=\"header\" patch={...} version=\"<branch>\"\n   update_page page_id=home patch={...} version=\"<branch>\"\n   ```\n\nb. Preview after every meaningful change:\n\n   ```\n   get_preview_link page_id=home version=\"<branch>\"\n   ```\n\n   Send the URL to the user. The preview navigates the whole branch\n   from one mint.\n\nc. Iterate on feedback. Common rounds: headline tightening, color\n   tweaks, swapping hero images.\n\n### 4. Site-wide changes through partials, not pages\n\nIf the redesign touches every page (e.g. new global header, new\nfooter, new CTA bar) — edit a partial, not 23 pages. Before editing a\nshared block, check the blast radius:\n\n```\nfind_pages_using_block partial_id=\"header\" version=\"<branch>\"\n```\n\nThis returns every page that would show the change. Communicate that\nto the user before the save.\n\n### 5. Bulk content cleanups via dry-run first\n\nIf the redesign requires content rewrites (e.g. \"remove every mention\nof the old company name\"), use the bulk tool with dry-run:\n\n```\nsearch_pages contains=\"OldCo\" version=\"<branch>\"\nbulk_replace_text pattern=\"OldCo\" replacement=\"NewCo\" dry_run=true version=\"<branch>\"\n# Show the user the sample_diffs\nbulk_replace_text pattern=\"OldCo\" replacement=\"NewCo\" dry_run=false version=\"<branch>\"\n```\n\n### 6. Approval round\n\n**Self-review is a multi-DIMENSION pass, not a glance — and most of it you\nMEASURE, not eyeball.** Structural checks (copy present, images return 200) are\nNOT a design review; never report \"approved\" off them. Don't just list the bugs\nyou happened to notice — walk every dimension below on the DB-live preview\n(a reused `get_preview_link`; browser tool + DOM reads), fix what you find,\nreload, re-check. No re-deploy between fixes — the preview renders from the DB.\n\n**Use `tr-design-review` (`read_skill tr-design-review`) for the HOW** — it has\nthe per-dimension measurement snippets (overflow ladder, computed contrast, touch\ntargets, the anti-lazy-load broken-image check) and the scorecard + verdict\nformat. The dimension summary below is the \"what\"; that skill is the runnable\nroutine. In particular: scroll-and-settle to trigger lazy images BEFORE any\nfull-page screenshot, or you'll report blank boxes that aren't real.\n\n1. **Responsive** — screenshot across a width ladder (mobile / tablet / laptop /\n   desktop / wide ≈390 / 768 / 1024 / 1440 / 1920px) AND sweep the page's own\n   `@media` breakpoints (read the page-scoped `<style>`; resize a few px below +\n   above each). At EVERY width: `document.documentElement.scrollWidth <=\n   clientWidth` (no horizontal scroll), grids flip cleanly, nothing squished /\n   orphaned / overlapping, no mid-word breaks. Two sizes is not enough — bugs hide\n   in between. Also check a short/landscape viewport and 200% browser zoom.\n2. **Visual & brand** — logo FULLY VISIBLE (not clipped by a header\n   `overflow:hidden` + overlap margin) and brand-compliant; screenshot the header\n   IN CONTEXT, never the logo element in isolation (that hides clipping).\n   Decoration robust at real size in the real browser: no hairline seam at a\n   divider (use `core/section` `divider_top`/`divider_bottom` — don't hand-roll a\n   band), no glow clipped to a hard edge, no `object-fit:cover` cropping faces, no\n   gradient fade-cutoff. Typography: body line-length ~45–75ch, consistent scale,\n   no awkward widows on headings. Palette adherence (no off-brand colours);\n   consistent spacing / alignment / radius / shadow.\n3. **Accessibility — MEASURE, don't eyeball** — compute actual contrast ratios\n   (WCAG AA: body ≥4.5:1, large/UI ≥3:1) and fix failures by deepening the\n   offending colour token; meaningful `alt` on every image; exactly one `<h1>` +\n   no skipped heading levels; visible `:focus-visible` on every interactive\n   element; every input has an associated `<label>`; touch targets ≥44px on\n   mobile; semantic landmarks (header/nav/main/footer) + nav `aria-label`; honour\n   `prefers-reduced-motion`.\n4. **Functional** — the form actually works (POST action correct, hidden token\n   non-empty, honeypot present + hidden; a long name/email doesn't break layout);\n   every link + in-page anchor resolves (each `#anchor` has a matching `id`; no\n   `href=\"#\"`/`\"\"`); ZERO console errors/warnings; interactions (menu toggle,\n   hover/focus/active) work.\n5. **Content** — no unrendered `{{…}}` tokens in the DOM; no placeholder/lorem;\n   copy still matches the source of truth (the live page) verbatim.\n6. **Findable (SEO/meta)** — `<title>` + meta description present + sensible;\n   `og:title`/`og:description`/`og:image`; canonical; favicon + apple-touch-icon;\n   `<html lang>`; `noindex` correct (branches must be noindex).\n7. **Fast (performance)** — images at sane sizes (not a 2048px file shown at\n   380px without a responsive variant), modern format (avif/webp), `width`/`height`\n   or aspect-ratio set (no layout shift), below-fold lazy / above-fold eager.\n8. **Cross-browser** — the same CSS renders differently per engine (the divider\n   seam was Chrome-only; WebKit/Firefox have their own). Re-check in another engine\n   if you can; if only Chromium is available, statically flag risky props\n   (`backdrop-filter` without fallback, `-webkit-`-only masks, `100vh` on mobile →\n   prefer `100svh`, `sticky` inside `overflow`).\n\nFix what you find and re-check before involving the user. \"Looks structurally\nfine\" ≠ \"looks good\", and \"looks good in Chrome at 1440\" ≠ \"works for everyone,\neverywhere\" — never report a design as approved/perfect off a glance or a partial\npass.\n\nThen give the user the **DB-live preview link** to review — and use the SAME\nlink for your own verification:\n\n- **While iterating (default):** a reused `get_preview_link` (mint once,\n  reuse — defaults to a 24h TTL). It renders from the database with NO build, so every\n  edit shows on reload, and one link navigates the whole branch (internal\n  links keep the token). The token URL is stable across edits — re-mint only\n  when the 24h lapses, never per edit. Do NOT deploy just to let the user (or\n  yourself) see a change.\n- **When they want the COMPILED static site** (a permanent bookmark, a\n  stakeholder link to the built output, or a final pre-merge check): deploy\n  the branch once (`trigger_deploy version=\"<branch>\"`) and share the stable\n  alias `https://<branch>.<project>.pages.dev` (the `<project>` is the part\n  after the hash in the returned `deploy_url`). Branch deploys are\n  `robots_blocked`, so it won't be indexed. The immutable per-deploy\n  `<hash>.pages.dev` is for your own one-off checks (a new hash each deploy).\n\nDefault: review/iterate on the reused DB-live preview; deploy only for the\ncompiled output or merge. Wait for an explicit \"looks good, ship it.\"\n\n### 7. Merge + deploy\n\n```\nmerge_branch version_id=\"<branch>\"        # branch's diffs land on main\ntrigger_deploy\nget_deploy_status job_id=<id>             # poll until succeeded\n```\n\nOptionally, after a successful deploy:\n\n```\ndelete_branch version_id=\"<branch>\"       # tidy up\n```\n\n(You can also leave the branch around as a record of the redesign;\ndisk cost is tiny.)\n\n## Pitfalls\n\n- **Forgetting `version=` on writes.** Every call you make on the\n  branch must include `version=<branch-id>`. A missing one writes\n  straight to main — silent and bad.\n- **Skipping discovery.** \"Modernize\" without first reading the site\n  produces a confidently-out-of-place result. Always sample existing\n  pages.\n- **Deploying to preview.** Don't `trigger_deploy` after every edit just to\n  see the change — that builds static pages and the URL is only as fresh as\n  the last build. Iterate on a reused DB-live `get_preview_link` (renders from\n  the DB, reflects edits on reload); deploy only for the compiled static\n  output or merge (steps 6–7).\n- **Auto-merge.** Don't `merge_branch` without explicit user sign-off.\n  Once merged, the only undo is another branch + reverse edits.\n- **Header rewrites that drop the brand block.** Even when the\n  redesign is dramatic, preserve the brand mark + the nav skeleton\n  unless the user said to redo them.\n\n## When to NOT use a branch\n\nTiny edits — \"fix the typo on the About page\" — don't need a branch.\nThe in-portal chat handles those directly on main. This skill is for\nwork where:\n\n- The user might want to walk away mid-redesign and come back later\n- Multiple changes need to ship together\n- The site is high-traffic and \"broken for an hour\" is unacceptable\n- Stakeholder review across multiple pages is expected\n\nIf none of those apply, edit main directly and move on.\n",
     "tr-responsive": "---\nname: tr-responsive\ndescription: Use when a layout must behave differently at different screen sizes — different grid columns per breakpoint, an icon-box that's icon-on-top on mobile but icon-left on tablet, hiding a block on small screens, fluid type. Triggers on \"responsive\", \"mobile/tablet/desktop layout\", \"stack on mobile\", \"X columns on desktop and Y on mobile\", \"olika på mobil/surfplatta\", \"responsivt\".\n---\n\n# Make a Typeroll block layout responsive\n\nTyperoll has a built-in five-breakpoint system. You almost never hand-write\nmedia queries — you set per-breakpoint values on responsive fields and the\nrenderer compiles the `@media` rules per block instance.\n\n## The five breakpoints (mobile-first)\n\n`mobile (<640) · tablet (≥640) · laptop (≥1024) · desktop (≥1280) · wide (≥1536)`\n\nA responsive field takes either a scalar (applies everywhere) or a sparse\nobject `{ mobile?, tablet?, laptop?, desktop?, wide? }`. Missing breakpoints\ninherit from the next smaller one. So you only set the breakpoints that change.\n\n## Setting per-breakpoint values\n\nUse `set_block_responsive` (or pass the object form directly in `add_block` /\n`update_block` data). `read_block_type <id>` tells you which fields are\n`responsive`.\n\n```\n# 4 columns on desktop, 2 on tablet, 1 on mobile:\nset_block_responsive target={kind:page,id:home} block_id=<grid-id>\n  field=cols value={ mobile: 1, tablet: 2, desktop: 4 }\n\n# icon-box: icon on top on phones, beside the text on tablet+:\nset_block_responsive ... block_id=<iconbox-id>\n  field=layout value={ mobile: \"icon-top\", tablet: \"icon-left\" }\n```\n\nPass a scalar to collapse a field back to one value everywhere.\n\n### Worked example — the classic feature grid\n\n\"4 cards/row with icon-on-top on desktop, 2/row with icon-left on a landscape\niPad, 1/row icon-on-top on a phone\":\n\n1. `core/grid` containing `core/icon_box` cards (or a `core/repeater` with\n   `item_block: core/icon_box` for a collection-driven list).\n2. On the grid: `cols = { mobile: 1, tablet: 2, desktop: 4 }`.\n3. On each icon_box (or the repeater's item defaults):\n   `layout = { mobile: \"icon-top\", tablet: \"icon-left\", desktop: \"icon-top\" }`.\n\nNo media queries authored — the build emits per-instance `@media` blocks and\nthe editor preview honours them. Flip the device toggle in the editor header\n(Mobil / Mobil-liggande / iPad / iPad-liggande / Desktop) to author and verify\neach breakpoint.\n\n## Hiding a block at some sizes\n\n`Block.hidden_on: Breakpoint[]` is universal — no per-block opt-in. E.g.\n`hidden_on: [\"mobile\"]` drops the block below 640px. Use it instead of building\na \"mobile-only\" duplicate.\n\n## Authoring a CUSTOM block type that's responsive\n\nTwo halves, BOTH required (`create_block_type` / `update_block_type`):\n\n1. Mark the field `responsive: true`.\n2. Expose it on the **outermost** template element as a CSS variable:\n   `style=\"--{field}:{{field}}\"`, then read `var(--{field})` in the block CSS.\n\nIf the field's value is directly usable CSS (e.g. `direction: row|column` →\n`flex-direction: var(--direction)`), you're done.\n\nIf it's a friendly **token** that maps to CSS (e.g. `layout: icon-left` →\n`flex-direction: row`), add a `responsive_css` map on the field — otherwise the\nper-breakpoint overrides silently do nothing (a `[style*=\"--field:token\"]`\nselector can't see a `@media` override):\n\n```\n{ name: \"layout\", type: \"select\", options: [\"icon-top\",\"icon-left\"],\n  default: \"icon-top\", responsive: true,\n  responsive_css: { \"icon-top\": \"--dir: column;\", \"icon-left\": \"--dir: row;\" } }\n```\n\nThen the block CSS reads `flex-direction: var(--dir, column)`.\n\n## Fluid type — usually automatic\n\n`core/heading` and prose already use `clamp()` to scale smoothly between mobile\nand desktop. `core/heading` separates semantic `level` (h1–h6, for SEO) from\nvisual `size` (sm–3xl/auto) — \"h1 but only as big as an h3\" is one field, no\nbreakpoints needed.\n\n## Gotchas\n\n- Setting a value only at `desktop` leaves smaller screens on the field\n  *default*, not on your value — set `mobile` too if you want a non-default\n  baseline (mobile-first).\n- The editor preview width is approximate on a narrow panel, but the breakpoint\n  you're editing is exact. Trust the deployed site / a wider window for `wide`.\n- **Never paper over horizontal overflow with `html,body{overflow-x:hidden}`.**\n  Setting `overflow-x:hidden` on `html` forces `overflow-y` to compute as `auto`\n  (CSS spec), turning `<html>` into a fixed-height nested scroller — the page\n  then won't scroll normally (`window.scrollY` sticks at 0) and renders blank\n  below the fold. Instead, find the element that overflows (a fixed width, a\n  `transform:rotate` card poking out, a decorative `::before`/`::after`, a grid\n  that didn't collapse) and fix THAT element's width / clip it with\n  `overflow:hidden` on its own section. Verify with\n  `document.documentElement.scrollWidth === clientWidth` at 360–390px.\n- **`core/grid` `stack_at` may not collapse on mobile** (a known platform bug):\n  the block writes `style=\"--cols:N\"` inline, and an inline custom property beats\n  the media query that tries to set `--cols:1`, so the grid stays N-up and text\n  wraps a letter per line. Workaround until fixed: in page-scoped CSS override the\n  real property, e.g. `@media(max-width:640px){.my-section [data-block=\"grid\"]{grid-template-columns:1fr!important}}`.\n- Background design reference: `docs/responsive-blocks.md` in the platform repo.\n",
     "tr-seo": "---\nname: tr-seo\ndescription: Use when the user asks to improve SEO, fix meta tags, add structured data, check page titles, or audit the site's search visibility. Triggers on \"SEO\", \"meta descriptions\", \"Google ranking\", \"structured data\", \"JSON-LD\", \"sitemap\", \"sökoptimering\", or \"hjälp mig synas på Google\".\n---\n\n# SEO audit and improvements for a Typeroll site\n\n> **The buffer model (draft writes).** Every content write in this recipe\n> (pages, blocks, partials, collection items) lands in an unsaved per-doc\n> DRAFT — deploys and plain previews only see SAVED content. For recipe-style\n> build work, pass `save: true` on write calls (the work is pre-approved by\n> the task itself), or run `commit_working_copy` per doc before any\n> `trigger_deploy`. Preview your drafts with `include_working_copy: true`.\n\n\n## What Typeroll handles automatically\n\n- `<html lang>` from site `language` setting (per-page override via `language` field)\n- `<title>` = `page.seo_title || page.title + settings.default_seo_suffix`\n- `<meta name=\"description\">` from `page.seo_description`\n- `<meta name=\"robots\">` from `page.noindex`\n- `<meta property=\"og:*\">` Open Graph tags from seo_title, seo_description, og_image\n- `<link rel=\"canonical\">` from `page.canonical_url` (falls back to the page's own URL)\n- Article schema from `kind: \"article\"` + `author` + `date_published`\n- Page schema from `kind: \"page\"` (default)\n- `robots.txt` from `settings.robots_txt`\n- Sanitized HTML that preserves semantic structure\n\n## Recipe\n\n### 1. Audit current state\n\n```\nlist_pages status=\"all\"\nread_site_settings\n```\n\nFor each page, check:\n- Is `seo_title` set? (if not, Google uses `title` + suffix — often fine)\n- Is `seo_description` set? (150–160 chars, unique per page, includes keywords)\n- Is `og_image` set for the homepage and key landing pages?\n- Does the page have exactly one `<h1>`?\n\n### 2. Fix missing meta descriptions\n\n```\nbatch_update_pages updates=[\n  {page_id: \"home\",     patch: {seo_description: \"Acme designar rum...\"}},\n  {page_id: \"om-oss\",   patch: {seo_description: \"Vi är ett...\"}},\n  {page_id: \"tjanster\", patch: {seo_description: \"Våra tjänster...\"}}\n]\n```\n\nGuidelines:\n- 150–160 characters\n- Include the most important keyword naturally\n- Make it a compelling reason to click, not a summary of the page's nav\n\n### 3. Fix page titles\n\nSEO title = what Google shows in search results.\n\nIf `settings.default_seo_suffix` is set (e.g. \" — Acme Studio\"), every\npage whose `seo_title` is empty will show `title + suffix`. That's usually\nfine for inner pages; set an explicit `seo_title` only when you want\nsomething different.\n\n```\nupdate_site_settings {\"default_seo_suffix\": \" — Acme Studio\"}\n\n# Set the canonical URL style once per site. Existing sites default to `always`.\nupdate_site_settings {\"trailing_slash\": \"always\"}\n\n# A page that must keep its exact campaign/title text can opt out.\nupdate_page {\"page_id\": \"campaign\", \"patch\": {\"append_seo_suffix\": false}, \"save\": true}\n\nupdate_page page_id=\"home\" patch={\n  \"seo_title\": \"Acme Studio — Inredningsdesign i Stockholm\"\n}\n```\n\n### 4. Add Open Graph images\n\nSet `og_image` on pages that get shared on social media. If the site has\na branded hero image, upload it:\n\n```\nupload_media_from_url url=\"https://...\" alt=\"Acme Studio — Inredningsdesign\"\n# → returns cdn_url\n\nbatch_update_pages updates=[\n  {page_id: \"home\",     patch: {og_image: \"<cdn_url>\"}},\n  {page_id: \"om-oss\",   patch: {og_image: \"<cdn_url>\"}}\n]\n```\n\nOG image dimensions: 1200×630px ideal. The platform doesn't resize —\nuse a correctly-sized source image.\n\n### 5. Add structured data (JSON-LD)\n\nTyperoll auto-generates Article and Page schema, but you can override or\nextend with custom JSON-LD per page. Example: LocalBusiness on the homepage.\n\n```\nupdate_page page_id=\"home\" patch={\n  \"json_ld\": \"{\\\"@context\\\":\\\"https://schema.org\\\",\\\"@type\\\":\\\"LocalBusiness\\\",\\\"name\\\":\\\"Acme Studio\\\",\\\"url\\\":\\\"https://acme.se\\\",\\\"telephone\\\":\\\"+46812345\\\",\\\"address\\\":{\\\"@type\\\":\\\"PostalAddress\\\",\\\"streetAddress\\\":\\\"Drottninggatan 1\\\",\\\"addressLocality\\\":\\\"Stockholm\\\",\\\"postalCode\\\":\\\"111 51\\\",\\\"addressCountry\\\":\\\"SE\\\"}}\"\n}\n```\n\n**Important:** JSON-LD goes in the `json_ld` field as a JSON *string*\n(not a nested object). The renderer injects it inside\n`<script type=\"application/ld+json\">`.\n\nCommon schemas worth adding:\n- Homepage: `LocalBusiness` or `Organization`\n- About: `AboutPage`\n- Contact: `ContactPage`\n- Blog articles: auto-generated from `kind:\"article\"` + `author`\n- Events: `Event` with `startDate`, `location`\n- Products: `Product` with `offers`\n\n### 6. robots.txt\n\nThe default robots.txt allows all crawlers. Update if needed:\n\n```\nupdate_site_settings {\n  \"robots_txt\": \"User-agent: *\\nAllow: /\\nSitemap: https://acme.se/sitemap.xml\"\n}\n```\n\nTyperoll doesn't generate a sitemap automatically in phase 1. If the\ncustomer needs one, create a `/sitemap` page with HTML that lists all\npublished pages, or write a static `sitemap.xml` as a page with\n`slug: \"sitemap.xml\"` and HTML-encoded XML (not recommended for large sites).\n\n### 7. Canonical URLs\n\nSet `canonical_url` when a page has a duplicate (e.g. the same content\naccessible via two slugs after a migration):\n\n```\nupdate_page page_id=\"tjansterna\" patch={\n  \"canonical_url\": \"https://acme.se/tjanster\",\n  \"noindex\": true\n}\n```\n\n### 8. Language settings\n\n```\nupdate_site_settings {\"language\": \"sv\"}\n```\n\nPer-page override for multilingual content:\n```\nupdate_page page_id=\"about-en\" patch={\"language\": \"en\"}\n```\n\n### 9. Heading audit\n\nUse `search_pages` to find structural problems:\n\n```\nsearch_pages contains=\"<h1\"      # pages that have at least one H1\n```\n\nThen `read_page` on pages that seem to have none or multiple. Fix via\n`update_page patch={html_content: \"<corrected HTML>\"}`.\n\n### 10. Deploy\n\n```\ntrigger_deploy\nget_deploy_status job_id=<id>\n```\n\n## Pitfalls\n\n- **Don't stuff keywords.** Write descriptions for humans. Google ignores\n  `<meta name=\"keywords\">` (not a field in Typeroll anyway).\n- **JSON-LD is a string, not a nested field.** Pass the entire schema as\n  a JSON-encoded string in `json_ld`. The server escapes `</script` before\n  injection.\n- **OG images need absolute URLs.** The `cdn.typeroll.com` URLs are always\n  absolute — use those.\n- **`canonical_url` + `noindex` together.** If you noindex a page AND set\n  canonical, the canonical is redundant (noindexed pages don't pass equity).\n  Use one or the other.\n- **Default suffix on homepage looks odd.** \"Acme Studio — Acme Studio\"\n  happens when title=\"Acme Studio\" and suffix=\" — Acme Studio\". Set an\n  explicit `seo_title` for the homepage.\n"
 };
 export const BUNDLED_DOCS = {
-    "agents": "# AGENTS.md — Working on a Typeroll site\n\nYou are connected to a Typeroll site through `@typeroll/mcp-server`.\nThis file is your briefing: what the system is, what conventions matter,\nwhat tools to reach for first.\n\nIf anything below conflicts with what you observe in the tools, trust the\ntools — the platform may have moved since this was written.\n\n**Start here for site-shaped tasks.** When the user wants to build,\nmigrate, redesign, or brand a site, call `list_skills` first — the server\nadvertises its own step-by-step playbook (`tr-new-site`, `tr-migrate-wp`,\n`tr-brand`, …). Then `read_skill name=…` loads the full recipe. These are\nlocal reads; no API key or site context required.\n\n**Branch first for anything larger than a small edit.** Before a redesign,\na multi-page change, or trying out a new design direction, run\n`create_branch name=\"…\"` and pass the returned id as `version=<id>` on every\nsubsequent read/write. The work stays off the live `main` version until you\n`merge_branch` it — nothing ships until you decide it should. Branches default\n`robots_blocked:true` and get their own deploy URL for stakeholder review.\nIt's the cheapest insurance there is; when in doubt, branch. The\n`tr-redesign-branch` skill walks the whole flow. (Small, low-risk single edits\ncan go straight to main.)\n\n## What this is\n\nTyperoll is a static-site CMS: content lives in a database, the user\nedits it through an in-app editor, and a deploy step compiles everything\nto a fast static site hosted on Cloudflare Pages. The in-app chat handles\nsingle-page or single-block edits by the editor audience. You — through\nthis MCP — handle the work that doesn't fit there: site-wide redesigns,\nbulk content updates, structural migrations, directory imports.\n\nThe MCP server is a thin wrapper around the public REST API. Each tool\nmaps to one HTTP endpoint; the actual logic runs in the customer's portal\n(SaaS or self-hosted).\n\n## The data model in 90 seconds\n\n- **Pages.** Title, slug, status (`draft | review | unlisted | published`),\n  body content + SEO fields. Two body shapes selectable per page via\n  `content_mode`:\n  - `blocks` (DEFAULT for new pages) — `blocks: Block[]` tree of typed\n    blocks (heading, prose, section, columns, image, button, plus any\n    user/third-party block types installed on the site). Use the\n    block-mutation tools (`add_block`, `update_block`, `move_block`,\n    `remove_block`) for structural changes.\n  - `html` — body lives in `html_content` as a single HTML string.\n    Useful when you have hand-written markup to drop in directly.\n\n  Slug is a single path segment — no slashes. `about` → `/about`,\n  `kontakt` → `/kontakt`, empty string `\"\"` → homepage. The v1 API\n  rejects `services/design` and other slash-containing slugs with\n  \"Invalid slug … slugs must not contain slashes.\" For nested URLs\n  like `/blog/{slug}` or `/services/{slug}`, the right primitive is a\n  **collection with `route_template`** (see the `tr-blog` and\n  `tr-directory` skills) — not a flat page with a slashed slug.\n\n- **Partials = global blocks.** Three kinds:\n  - `header` — auto-injected at the top of every page.\n  - `footer` — auto-injected at the bottom of every page.\n  - `free` — reusable HTML you drop into a page with\n    `<x-include name=\"block-id\" />`. Free blocks are how you avoid\n    duplicating HTML across HTML-mode pages.\n\n  Partials themselves also support `content_mode='blocks'` — pass a\n  `blocks: Block[]` tree to `update_partial` and the renderer composes\n  it the same way as a page. Useful for header/footer authored with\n  block types.\n\n- **Collections.** Repeatable content types (blog, team, events,\n  products, restaurants for a directory site, etc.). Each has a schema\n  (`fields[]`) and optional **per-item routing** via `route_template`\n  (e.g. `/restaurants/{slug}`). When set, every published item gets its\n  own static URL rendered through `item_template_html`. Set\n  `route_template=\"\"` to opt out and keep the collection listing-only.\n\n- **Settings.** Site name, tagline, logo, favicon, colors, fonts,\n  contact info, social links, SEO suffix, default meta description\n  (`default_meta_description` — site-wide fallback for pages without a\n  `seo_description`; tagline is the last resort), plus `scripts_head`,\n  `scripts_body_end`, `custom_css` (writable via the API — your bearer\n  token authorises shipping arbitrary CSS/JS to the live site, just\n  like editing a partial's HTML does).\n\n- **Analytics attribution.** `read_funnel_attribution` and\n  `update_funnel_attribution` manage the Analytics module's validated,\n  site-level rules that forward\n  allowlisted query parameters to exact HTTPS link targets. Optional\n  first-/last-touch storage is consent-gated and off unless configured. Read\n  the current config before writing. For advertising pass-through, preserve\n  only incoming or stored values and omit `fallback`: a fallback creates\n  synthetic attribution when no campaign value exists. The API rejects such\n  fallbacks unless `allow_synthetic_fallbacks=true` explicitly acknowledges\n  that behavior. A target `click_event` is recorded as a consent-gated,\n  first-party conversion when Analytics is enabled; the event contains only\n  the declared funnel, destination, page path, and resolved allowlisted\n  attribution. Navigation never waits for analytics.\n\n- **Core modules.** `list_apps`, `read_app`, and `update_app` expose the\n  code-defined core-module registry (the `apps` API name is retained for\n  compatibility) through the same admin API key used for content\n  and deploys. Read the schema before writing. Secret fields stay masked on\n  reads and encrypted at rest; omitted fields preserve their current values.\n  When `affects_build` is true, deploy after the update.\n\n- **Extension installations.** `list_extension_installations`,\n  `read_extension_installation`, and `update_extension_installation_config`\n  expose each installed Extension's manifest-defined config through the admin\n  API key. Read the installation before writing so you use the exact keys from\n  `manifest.config_schema`. This is the supported automation path for public\n  content such as consent copy, link text, and policy URLs; masked secrets are\n  preserved when omitted. A successful update returns\n  `redeploy_required: true`; review the result, then deploy explicitly.\n\n- **Page templates.** A `PageTemplate` is a Block[] tree that wraps a\n  page's body. The template contains exactly one block of type\n  `template_content_slot` — at render time that block gets replaced by\n  the page's own `blocks`. Set `Page.template = \"<template-id>\"` to\n  apply a template to a page.\n\n- **Block types.** A site has three sources of block types:\n  - **Core** (origin: 'core', ids like `core/section`) — shipped in\n    the platform, always available.\n  - **User** (origin: 'user') — created in the portal's block-types UI.\n  - **Third-party** (origin: 'third_party') — imported from .tcblocks\n    packages via `import_block_types`.\n\n  `list_block_types` returns ALL of them in one list as a lightweight\n  summary: each entry's id, label, category, container/slot info, origin,\n  and full field schema (names, types, defaults) — but NOT the render-time\n  template/styles/script (omitted so the list stays within token budget as\n  the library grows). Use `read_block_type` for one block's markup, or pass\n  `full:true` to inline it for every block. Always call this FIRST before\n  working with blocks — never hardcode block ids or field names, the\n  available set is per-site.\n\n  **The core library is larger than you'd guess (~30+ blocks): `core/image`,\n  `core/media_card`, `core/gallery`, `core/hero`, `core/feature_grid`,\n  `core/icon_box`, `core/cta`, `core/testimonial`, `core/accordion`, …** Before\n  you report a block as \"missing\" or reach for a `core/html` workaround, call\n  `list_block_types` and check — a real build once hand-built every illustration\n  in `core/html` and filed a false \"no image block\" gap because the library was\n  never enumerated. Prefer a native block; `core/html` is the last resort.\n\n  Block-library specifics worth knowing (template_capabilities_version\n  0.15.0):\n  - **`core/media_card`** — image + text side by side (image left/right,\n    width third/two-fifths/half, heading + richtext + button, optional\n    card background/radius; stacks image-on-top below 720px). Use it for\n    the classic \"photo next to copy\" layout instead of hand-building\n    section+grid+html.\n  - **`core/search`** (0.29.0+) — site search over the deployed site.\n    Place the block anywhere; the deploy pipeline detects it, runs\n    Pagefind over the built HTML, and the block loads the search UI at\n    visit time. Index = page content only (nav/footer excluded); noindex\n    pages stay out. Editor preview shows a placeholder note (the index\n    only exists on the deployed site).\n  - **Archive pagination** (0.29.0+): a collection listing\n    (`core/collection_list` / `core/repeater`) with `paginate: N` renders\n    N items per page + a pager, and the build generates `/page/2/`… routes\n    automatically. `paginate` supersedes `limit`; one paginated listing\n    per page.\n  - **`core/feature_row`** (0.29.0+) — the full-width \"zig-zag\"\n    feature/step row: balanced halves that hug the center gutter (no\n    wide-screen dead air), natural-aspect image (never cover-cropped —\n    that's media_card's card look), eyebrow + heading + richtext +\n    button pair, `image_side: left|right` per row, `stack_order`\n    controls what comes first on mobile. Prefer it over `core/columns`\n    with an unbalanced ratio + width-capped text for these rows.\n  - **`core/hero` and `core/cta` render their buttons server-side** via\n    `primary_label`/`primary_url` + `secondary_label`/`secondary_url`.\n    (The old `buttons` array relied on client hydration that never\n    existed — if you see `data-buttons` in stored content it renders\n    nothing; rebuild with the explicit fields.)\n  - **`core/image` gets responsive `<picture>` automatically at build\n    time** — the deploy pipeline's SEO transform converts CDN `<img>`\n    into `<picture>` with AVIF/WebP srcset variants. You do NOT need\n    `core/html` for responsive images; just point `src` at an uploaded\n    media URL (run `generate_image_variants` first) and optionally set\n    `radius`. Note: the in-portal preview shows the plain `<img>` — the\n    `<picture>` upgrade appears on the deployed site.\n  - **Icons render inline SVG** (since template_capabilities_version\n    0.16.0). Every `type: 'icon'` schema field — on `core/icon`,\n    `core/icon_box`, `core/step_card`, and custom block types — renders\n    a stroke-based inline SVG when the value is a name from\n    `get_site_capabilities → core_icon_names` (a curated Lucide subset:\n    `check`, `star`, `shield-check`, `mail`, `arrow-right`, `zap`,\n    `truck`, `chart-line`, …). Any other value (emoji, plain text) is\n    rendered as escaped text, so emoji stand-ins keep working. Icons\n    size with `font-size` (the SVG is 1em) and paint with\n    `currentColor`. Custom block templates opt in by placing the derived\n    raw token `{{{<field>_svg}}}` where the icon should appear. On\n    pre-0.16.0 portals icons don't render — use emoji or CSS markers.\n    `core/tabs` label icons are the remaining gap (tab strip is built\n    client-side).\n  - **Grids with a partial last row: set `last_row: 'center'`** (since\n    template_capabilities_version 0.16.5). Five equal cards in a 3-col\n    `core/grid` (or 7 in 4, ...) left-align the orphans by default; with\n    `last_row: 'center'` the last row auto-centers. THE DESIGN RULE: when\n    N peer cards don't divide by the column count, center the last row or\n    change the column count — NEVER invent a \"wide\"/full-width variant of\n    one peer card just to fill the hole. Special treatment is a content\n    decision, not a layout patch.\n  - **`core/section` is natively full-bleed on block pages** (since\n    template_capabilities_version 0.14.0): the section's background runs\n    edge-to-edge and meets the header with zero gap; content inside is\n    constrained by the section's own inner container (`width` field:\n    narrow/normal/wide/full). Never use 100vw negative-margin hacks.\n    Top-level blocks that are NOT sections still get a classic centered\n    container as fallback. Anchor ids and custom classes via\n    `style_overrides` are safe on full-bleed sections since 0.15.3 —\n    they merge into the `<section>` element itself. On 0.14.x–0.15.2\n    they wrapped the section in a `<div>`, which silently disabled\n    full-bleed for that section.\n  - **Shaped section transitions** (since template_capabilities_version\n    0.24.0): `core/section` takes `divider_top` / `divider_bottom`\n    (`none | wave | curve | tilt`). The platform paints the divider in the\n    section's OWN `background` and overlaps the neighbour by 1px, so a\n    cream↔colour transition renders seam-free. **Use this for waves/curves —\n    never hand-roll a divider band in `core/html`** (a separate stacked shape\n    seams against the next section as a sub-pixel hairline in Chrome). Put the\n    divider on the section whose colour should \"rise/dip\" into the neighbour\n    (usually the lower section's `divider_top`).\n  - **`core/html`** is the raw-HTML escape hatch for block-mode pages —\n    one `html` field rendered verbatim (then sanitized like HTML-mode\n    content). Use it for the genuinely unique thing no block covers.\n    Prefer real blocks when one fits.\n  - **Structured records at scale** (template_capabilities_version ≥ 0.31.0).\n    Four things landed together for directory-shaped sites:\n    - `collection_completeness` — **start an enrichment pass here**, not by\n      paging every item. Returns per-field gap counts plus the N worst\n      records (missing fields, never-verified fields, fields whose last write\n      is older than the staleness window), computed at read time. Fields no\n      API key may write are excluded by default: a gap you can't close is\n      noise.\n    - **Per-field write authority.** A collection field can declare\n      `writable_by` (`portal | owner | agent | app | import`). A write you're\n      not permitted, or one that would overwrite a higher-precedence writer\n      (a human correction, the listed business's own edit), comes back as\n      **409 with the losing field names** — never a silent no-op. Treat that\n      as \"already handled\" and record it; retrying will lose again.\n    - **Item references.** `item_ref` / `item_ref_list` fields point at items\n      in another collection (`ref_collection`). The reverse direction is\n      computed at render time — don't try to maintain backlinks yourself.\n      Render them with a `core/repeater` whose `source_type` is `related`\n      (a ref field on the current item) or `backlinks` (who points at it).\n    - **Taxonomy pages.** `CollectionDef.facets` generates one page per\n      distinct field value. ⚠️ This turns record count into ROUTE count, and\n      route count is what the build timeout measures. `min_items` (default 2)\n      keeps thin-content pages out, and combination pages must be listed\n      explicitly in `facet_combinations` — never assume a cartesian product.\n  - **`core/embed`** (template_capabilities_version ≥ 0.30.0) is\n    `core/html` plus behaviour: an `html` field and a `js` field. Reach\n    for it when a one-off placement needs JavaScript. **A `<script>` tag\n    written into `core/html` — or into any page/block markup — is stripped\n    by the sanitizer no matter which credential wrote it**, so this field\n    is the supported route, not a workaround. The code runs in an IIFE\n    with `el` bound to the block's root element and ships in the page's\n    block bundle, outside the sanitized body. Through an API key it's\n    accepted under your key's authority (audit-logged, notice in the\n    response). Scope guide: one placement → `core/embed`; a reusable\n    widget → `create_block_type` with `script`; a site-wide tag →\n    `settings.scripts_head` / `scripts_body_end`.\n  - **Forms 2.0** (template_capabilities_version ≥ 0.18.0): forms can\n    carry `steps[]` — each step is a Block[] tree mixing `form/*` field\n    blocks (text/email/phone/number, textarea, select/radio_group/\n    checkbox_group, toggle, slider, date, URL, heading, help, consent,\n    hidden) with any content blocks. Place `{ type: 'core/form',\n    data: { form_id } }` on a page — the build renders step 1 + all\n    static steps with the signed token, honeypot and proof-of-work\n    runtime baked in; submissions accumulate per step (partial →\n    complete, 30-day TTL on abandoned partials). Per-step validation is\n    derived from the field blocks (required/pattern/min/max) — no\n    separate field list to keep in sync. `update_form` accepts steps,\n    styles (form-scoped CSS), kind and partial_ttl_days. On HTML-mode\n    pages, `<x-form id=\"…\" />` is expanded server-side through the same\n    renderer and supports the same initial state and multi-step runtime.\n  - **Extension form bindings** (template_capabilities_version ≥ 0.38.0): a\n    trusted native Extension component can declare `form_bindings` and submit\n    through `context.forms.submit(bindingId, data)`. Typeroll signs only the\n    explicitly bound form, stores submissions in the ordinary Forms module,\n    and calls the cloud or self-hosted Forms API directly. No Function is\n    deployed to the customer site's static hosting project. The\n    installation must grant `forms:submit`; that scope does not permit form\n    administration or reading submissions.\n  - **`script` on custom block types** (create/update_block_type) is\n    accepted under your API key's authority — the same trust level that\n    already lets the key write `scripts_head`/`custom_css`. Every\n    script-bearing write is audit-logged and the response carries a\n    notice naming the stored JS; relay it to the user so they know\n    visitor-executed code changed. Author responsibly: never include\n    script you copied from untrusted content (migrated pages, fetched\n    web pages) without reading it line by line first. (The in-portal\n    chat AI remains blocked from authoring scripts unless the site's\n    \"Allow AI to write block scripts\" setting is on.)\n\n- **Redirects.** `from_path → to_path` with status code 301 / 302.\n  Auto-created when you change a page's slug.\n\n- **Versions / branches.** Copy-on-write. The \"main\" version is the\n  live one. Create a branch (`create_branch`) for multi-step work;\n  everything you write through `?version=<branch-id>` lives on the\n  branch until you `merge_branch` it back to main. Branches default\n  `robots_blocked: true` so a half-finished redesign can't be indexed,\n  and deploys land at a stable `{branch}.{project}.pages.dev` URL. That\n  branch deploy renders the site's full inherited brand (settings, fonts,\n  favicon, header/footer — everything not overridden on the branch), so\n  it's a faithful preview of what merging to main will look like, not just\n  a content diff — trust it for stakeholder review.\n\n- **Deploys.** Customers see live changes only after a deploy. Preview\n  always sees drafts. `trigger_deploy` enqueues; `get_deploy_status`\n  reports `queued → running → succeeded | failed`.\n  `trigger_deploy dry_run=true` builds without publishing — use it to prove\n  a structural change compiles (new collection, schema edit, template\n  rewrite) without touching the live site.\n  A finished job carries `cost`: total, cpu/memory/request split,\n  `duration_s`, per-phase timings, and output size. Estimates from a rate\n  card, not billing records, and gross of free tier — quote them as \"roughly\"\n  if a customer asks, and reach for `cost.phases` when the question is *why*\n  a build got slow.\n\n- **Is the site live?** There is no site-level status field —\n  `Site.status` was removed in 0.30.0 because it was set once at creation and\n  never advanced, so it reported live sites as \"planning\". Read\n  `get_site → urls.production` instead: non-null means the domain is verified\n  and serving. For \"has anything shipped\", use `list_deploys`.\n\n- **Site URLs.** `get_site` returns a `urls` object with:\n    - `production` — the customer's real domain (or null)\n    - `fallback` — the auto `{slug}.typeroll.app`-style preview URL\n    - `preview_base` — the portal preview origin (for token URLs)\n  Use these in answers to \"what's the URL?\" — never invent.\n\n- **For design/content iteration, share the DB-LIVE preview — don't deploy.**\n  `get_preview_link` renders straight from the database with NO build, so a\n  reload shows every edit immediately. Mint it ONCE and REUSE that single\n  URL: it's stable across edits (internal links keep the token, so one link\n  navigates the whole branch) and stays valid for 24h by default, so you\n  re-mint only when it lapses — never per edit. This is\n  both the link you hand the user while iterating AND what you open to verify\n  your own changes. Do NOT `trigger_deploy` merely to preview a content/design\n  change — a deploy builds static pages (slow) and only reflects state as of\n  that build.\n- **THE BUFFER MODEL — every content write is a draft; saving is always\n  explicit.** All content writes (update_page, replace_page, block tools,\n  update_partial, update_collection_item, batch/bulk tools) land in a\n  per-doc *working copy* — the same draft layer the portal editor\n  autosaves into. Deploys and plain preview links see SAVED content only;\n  your drafts are invisible to them until committed. The loop:\n    1. Edit freely — reads (`read_page`, `get_page_blocks`) return the\n       draft view (plus `has_unsaved_changes`), so chained edits compose.\n    2. Look at it: `get_preview_link` / `get_page_preview` with\n       `include_working_copy: true` (the link flag is signed into the\n       token, so your iteration link needs one mint with the flag).\n    3. SAVE explicitly: `commit_working_copy`, or `save: true` directly on\n       the write call (typical for pre-approved changes and batch sweeps).\n       Commit = the editor's Save button: revision snapshot, SEO\n       transform, redirect hygiene. Rejected → `discard_working_copy`.\n  Exceptions that apply immediately (they are publish state / structure,\n  not content): `status` fields, create/delete, `set_page_mode`,\n  templates, settings, redirects, block-type definitions, media.\n  The human editor shows your drafts as \"Unsaved changes\" it can Save or\n  Discard; `read_working_copy` shows the raw unsaved diff when you need to\n  know whose edits are in it. Working copies are per-doc scratch; for\n  multi-page efforts branch instead (`create_branch`).\n  **Before `trigger_deploy`: commit.** Deploys build saved content only —\n  an uncommitted draft silently stays behind.\n- **Deploys / `{branch}.{project}.pages.dev` are the STATIC BUILD**, refreshed\n  only by `trigger_deploy`. Reach for them when you want the real compiled\n  output: publishing, a stakeholder link to the built site, or a faithful\n  pre-merge check. The branch alias is permanent across re-deploys; the\n  per-deploy `{hash}.pages.dev` is immutable per build. Reserve deploys for\n  these — not for previewing edits.\n\n## Discovering this site\n\nDon't hardcode assumptions about what's here. Every fact about the site\ngoes through the MCP:\n\n**Capability discovery is a required gate for every build, migration and\nredesign.** Before choosing HTML mode, hand-writing a component, or reporting\nthat Typeroll lacks a feature, call `get_site_capabilities` and\n`list_block_types` (`full:true` only when you need every template). If a likely\ntype appears, call `read_block_type` and inspect its schema. A capability gap is\nvalid only after those reads show that neither a core/site block nor a\ncomposition of `core/section`, layout blocks, `core/repeater`, template blocks,\nor a custom block type can express the requirement. Record the calls and the\nclosest available primitive in any gap report. This is a completion criterion,\nnot optional discovery.\n\n1. `get_site` — confirm the key works; learn the site name + URLs.\n2. `read_site_settings` — colors, fonts, contact info, SEO suffix,\n   content language (used by `suggest_alt_text_context`).\n3. `list_pages` — what pages exist, paginated.\n4. `list_partials` — what shared blocks already exist. **Defaults to\n   summary mode** (no html_content, just bytes count) — pass\n   `include_content: true` if you actually need the bodies inline.\n5. `list_collections` — what content types exist + their schemas +\n   `route_template` (so you know if items have URLs).\n6. `get_site_capabilities` — renderer version and feature flags. Never infer\n   support from remembered release notes.\n7. `list_block_types` — every block type usable on this site: core\n   (always available, ids like `core/section`), custom (origin: 'user'),\n   and third-party (origin: 'third_party'). Each entry includes the\n   full schema so you know what `data.X` fields each block accepts.\n8. `list_page_templates` — PageTemplate docs that wrap pages.\n\nFor a build, migration, redesign, or capability report, #1–#8 are the\npreflight. For a small content-only edit, #1 + #2 + a sampling from #3 is\nusually sufficient.\n\n**Source of truth = the live site (the API), by default.** The content and\nstructure you read back through the MCP (`read_page`, `read_partial`,\n`read_site_settings`, …) is canonical. Local files in the project folder —\n`sources/*.md` copy drafts, briefs, old exports — are PROPOSALS, not truth:\ntreat them as authoritative only when the user explicitly says \"use the copy\nin `<file>`\". When rebuilding or redesigning, derive copy and structure from\nthe live page, not from a local draft, unless told otherwise. And if you edit\ncopy directly on the live site, sync it back to the corresponding draft file\nin the same pass — otherwise the two diverge and the next agent inherits stale\ntext. (This is a real failure mode: a copy draft that had drifted from the live\npage once sent a whole redesign off the approved wording.)\n\n**Don't have a site yet?** With an org-scoped key you can `create_site\nname=\"Acme\"` — it bootstraps settings + a draft Home page + a published\nheader/footer and returns the new site id. Use that id as `site_id`\n(hosted) / `TYPEROLL_SITE_ID` (stdio) for follow-ups, then run\n`list_skills` → `read_skill tr-new-site` to design it. A site-scoped key\ncan't create sites (it's bound to one) and gets a 403.\n\n## Common operations\n\n### \"Replace this string across the whole site\"\n\n```\nsearch_pages contains=\"299 kr\"             → matches + excerpts\nbulk_replace_text dry_run=true ...          → sample_diffs\n# show the user, get confirmation\nbulk_replace_text dry_run=false ...         → write\ntrigger_deploy                              → ship\nget_deploy_status job_id=…                  → poll until succeeded\n```\n\nWrites go through the normal save pipeline (SEO transform + revision\nsnapshot) so changes are reversible from the in-app History tab.\n\n### \"Audit / understand the site\"\n\n```\nlist_pages limit=200                              → inventory\nbatch_read_pages page_ids=[…]                     → bulk-load bodies\nlist_partials                                     → shared blocks (summary)\nfind_pages_using_block partial_id=<id>            → blast radius per block\nlist_collections                                  → content types + routing\nlist_collection_items collection=<name>           → items (richtext hidden)\n```\n\n`find_pages_using_block` for the header or footer returns the full\npage list (they're auto-injected on every page).\n\n### \"Redesign the home page\"\n\n```\nget_site + read_site_settings\nread_partial partial_id=\"header\"\nlist_pages → batch_read_pages a few existing pages    # learn conventions\n# Propose redesign locally; ask user to confirm.\ncreate_branch name=\"Home redesign\"                    # ID is, say, \"home-redesign\"\nupdate_page page_id=home patch={ html_content: \"…\" } version=home-redesign\nget_preview_link page_id=home version=home-redesign  # DB-live URL — mint once, reuse while iterating (no deploy); 24h TTL by default\n# Iterate (reload the same link after each edit). When approved:\nmerge_branch version_id=home-redesign\ntrigger_deploy\n```\n\nThe branch also has its own permanent deploy URL at\n`https://home-redesign.<project>.pages.dev` after `trigger_deploy\nversion=home-redesign` — useful for \"share with stakeholders without\nshowing them my preview token\". `read_version version_id=home-redesign`\nreturns it as `deploy_url`.\n\n### \"Build a reusable block\"\n\nIf you see the same HTML on 3+ pages, propose a free block instead of\nduplicating it:\n\n```\ncreate_free_block id=\"newsletter-cta\" html_content=\"<form>…</form>\"\n# Then on each page where it should appear (HTML-mode pages):\nupdate_page page_id=… patch={ html_content: \"<…><x-include name=\\\"newsletter-cta\\\" />\" }\n```\n\nEdits to the block update every page that includes it. Use\n`find_pages_using_block` before changing it.\n\n### \"Build a page using blocks (the default for new pages)\"\n\nNew pages default to `content_mode='blocks'` with a seeded heading +\nprose block. Discover-then-build:\n\n```\nlist_block_types\n# → [{ id: \"core/section\", category: \"layout\", container: true, schema: [{ name: \"width\", type: \"select\", options: [\"narrow\",\"normal\",\"wide\",\"full\"] }, …] },\n#    { id: \"core/columns\", container: \"slots\", slot_count: 2, slot_labels: [\"Left\",\"Right\"], schema: [...] },\n#    { id: \"hero_bold\", origin: \"user\", schema: [...] },   ← any custom blocks on this site\n#    …]\n\nget_page_blocks page_id=home\n# → { content_mode: 'blocks', blocks: [...] }\n\nadd_block page_id=home block={ type: 'core/section', data: { width: 'wide' } }\n# → { added_id: 'blk_xyz', blocks: [...] }\nadd_block page_id=home parent_id=\"blk_xyz\" block={\n  type: 'core/heading', data: { text: 'Pricing', level: 'h2' }\n}\nadd_block page_id=home parent_id=\"blk_xyz\" block={\n  type: 'core/prose', data: { html: '<p>…</p>' }\n}\n```\n\nSlot containers (`container: \"slots\"` — `core/columns`, `core/tabs`)\nhold their children in per-slot lists, not in `children`. Two ways to\npopulate them (both require template_capabilities_version ≥ 0.15.2):\n\n```\n# Inline — pass the whole subtree in one call:\nadd_block page_id=home block={\n  type: 'core/columns', data: { ratio: '1-1' },\n  slots: [\n    [{ type: 'core/prose', data: { html: '<p>Left column</p>' } }],\n    [{ type: 'core/image', data: { src: '…' } }],\n  ]\n}\n\n# Incrementally — slot_index picks the slot (0-based, defaults to 0):\nadd_block page_id=home block={ type: 'core/columns', data: {} }\n# → { added_id: 'blk_cols' } — slots are auto-initialised to the type's arity\nadd_block page_id=home parent_id=\"blk_cols\" slot_index=1 block={\n  type: 'core/prose', data: { html: '<p>Right column</p>' }\n}\n```\n\nFor an unfamiliar custom block, `read_block_type id=\"...\"` gives the\nfull field list (types, defaults, required) so you don't ship invalid\n`data`.\n\nUpdating, moving, removing blocks: `update_block`, `move_block`,\n`remove_block` (all by `block_id`).\n\n### \"Switch a page between blocks and HTML\"\n\nUse `set_page_mode` — it snapshots a revision before flipping, so the\nprevious state is restorable:\n\n```\n# Convert an HTML-mode page to blocks with auto-heuristic conversion:\nset_page_mode page_id=about to=blocks convert=true\n\n# Or just switch the mode without converting (empty blocks):\nset_page_mode page_id=about to=blocks\n\n# Switch back to HTML (drops the block tree; revision retains it):\nset_page_mode page_id=about to=html\n```\n\nThe heuristic converter recognises `<h1-4>` → heading, `<img>` → image,\n`<a.btn>` → button, `grid-cols-2` → two-column, `<section>` / hero divs\n→ section. Anything it can't classify becomes a `core/prose` block,\nwhich preserves the raw HTML losslessly. Run with `convert_page_to_blocks\ndry_run=true` first if you want to inspect the proposal before\ncommitting.\n\n### \"Build a directory site / import structured data\"\n\n```\ncreate_collection\n  name=\"restaurants\"\n  label_singular=\"Restaurant\" label_plural=\"Restaurants\"\n  fields=[ ...title, slug, address, phone, cuisine, body... ]\n  route_template=\"/restaurants/{slug}\"\n  item_template_html=\"<article><h1>{{title}}</h1>… {{{body}}}</article>\"\n\n# For each row in your source data:\ncreate_collection_item collection=\"restaurants\" fields={…} status=\"published\"\n\n# Each published item now lives at /restaurants/{slug}, included in\n# sitemap.xml. Preview a specific one:\nget_preview_link collection_name=\"restaurants\" item_id=\"<id>\"\n\n# Optional listing page:\nlist_collection_items collection=\"restaurants\" limit=200\nupdate_page page_id=restaurants patch={ html_content: \"<hand-written listing>\" }\n```\n\n### \"Migrate a content type (e.g. WP custom post type)\"\n\n```\nlist_collections                                     # what exists today?\nread_collection name=blog                            # what fields are writable?\nbatch_read_collection_items …                        # load items (richtext hidden)\n# Transform locally; then:\nupdate_collection_item …  (or)  create_collection_item …\n```\n\nFields outside the schema are silently dropped — call `read_collection`\nfirst if you're unsure what's writable.\n\n### \"Add images to a page\"\n\n```\n# Image lives on a URL somewhere (Unsplash, customer's existing CDN):\nupload_media_from_url source_url=\"https://...\" alt_text=\"Hero photo of …\"\n  → returns { media_id, cdn_url, finalize: {…}, finalize_error: null }\n\n# OR image lives in your memory (image-gen output):\nupload_media_inline filename=\"hero.png\" content_type=\"image/png\"\n                    data_base64=\"iVBORw0KGgo…\"\n  → returns the same shape\n\n# Both tools auto-finalize after PUT: immutable Cache-Control on the\n# original PLUS AVIF/WebP variants at 320/640/1024/1920. No manual\n# generate_image_variants call needed. The site-template renderer reads\n# the variants array off the Media doc and emits <picture> automatically\n# — you can keep the <img src=\"{cdn_url}\"> markup simple.\n#\n# INTEGRITY — don't lose bytes in transit. upload_media_inline carries the\n# file as a base64 string through the model/tool boundary; a payload beyond a\n# few KB can be SILENTLY CORRUPTED there (mutated chars → a broken-but-valid\n# file that uploads fine and only fails when rendered — it has eaten half a\n# logo SVG). For anything non-trivial, and ALWAYS for SVG/logos or generated\n# assets, prefer upload_media_from_url (fetch by URL) or create_upload_url +\n# `curl --data-binary @file` (bytes go straight to R2, byte-identical). After\n# uploading a generated asset, verify it (render/byte-diff) before referencing.\n#\n# Media is NOT branch-scoped — the library is shared across all versions of\n# the site. Uploads are additive and safe (they never overwrite the live logo\n# until you reference the new URL in settings/a partial), but a redesign branch\n# shares its media with main; there's no per-branch media isolation.\n\n# Then embed in a page:\nread_page page_id=...\nupdate_page page_id=... patch={ html_content: \"<...><img src='{cdn_url}' alt='…' /></...>\" }\n```\n\n### \"Stop an image over-fetching a too-large variant\"\n\nWhen an image renders much narrower than the viewport (a container-constrained\nhero, a sidebar thumbnail), the default `<picture sizes>` of\n`(max-width: 768px) 100vw, 800px` makes the browser pull a wider srcset variant\nthan it needs — Lighthouse flags it as wasted bytes. Three levers, narrowest\nwins:\n\n```\n# 1. Per-image: put a real `sizes` on the <img>. Survives the transform verbatim.\nupdate_page page_id=... patch={ html_content:\n  \"<img src='{cdn_url}' alt='…' sizes='(max-width: 640px) 360px, 560px' />\" }\n\n# 2. Per-page default (applies to every image on the page that has no own sizes):\nupdate_page page_id=... patch={ image_sizes_default: \"(max-width: 640px) 360px, 560px\" }\n\n# 3. Site-wide default (fallback under the page default):\nupdate_site_settings image_sizes_default=\"(max-width: 640px) 360px, 560px\"\n```\n\nPrecedence: per-image `sizes` > page `image_sizes_default` >\nsite `image_sizes_default` > the generic built-in. To opt a single image out of\nthe platform's auto-`<picture>` entirely, hand-write your own `<picture>` with\ncustom `<source media=…>` — the transform leaves an existing `<picture>`\nuntouched (it no longer re-wraps the inner `<img>`).\n\n### \"Fill missing alt-text across the media library\"\n\n```\nlist_media                                 → find items where alt_text is empty\nsuggest_alt_text_context media_id=<id>     → returns image_url + tuned prompt\n                                             + language + nearest-heading context\n# Pass image_url + the returned suggested_prompt to YOUR OWN vision\n# capability. The platform does NOT run vision for you.\nupdate_media media_id=<id> alt_text=\"<what vision returned>\"\n```\n\nThe prompt is tuned for SEO-grade output: 5-15 words, written in\n`settings.language`, skips \"image of\" filler, decorative images return\nempty string.\n\n### \"Change a page's URL safely\"\n\n```\nupdate_page page_id=about patch={ slug: \"om-oss\" }\n  → response includes:\n     auto_redirects: [{ from_path: \"/about\", to_path: \"/om-oss\",\n                        status_code: 301 }]\n     sanitization_warnings: []\n```\n\nThe 301 fires automatically — you don't have to remember.\n\nRedirect hygiene is automatic in both directions (since 0.16.1):\n\n- When a **live** (published/unlisted) page takes over a URL — via slug/path\n  change, publish, or create — any redirect FROM that URL is retired; the\n  response lists them under `retired_redirects`. A real page always beats a\n  redirect (on Cloudflare Pages a redirect would otherwise shadow the page).\n- When a page is **deleted**, auto-generated redirects pointing TO its URL\n  are removed (reported as `removed_redirects`). Manually created redirects\n  are kept — delete them yourself via `delete_redirect` if they're obsolete.\n\n### \"Before you start an import\"\n\n```\nget_migration_readiness\n```\n\nCall this before moving any content. Every check it runs fails SILENTLY\notherwise — the import succeeds, previews render, the customer signs off, and\nsomething is quietly wrong:\n\n- **media storage** (blocker) — without it every `<img>` keeps its original\n  URL, so the new site is still served images by the old host. Nothing looks\n  broken until that hosting is cancelled, at which point every image on every\n  page breaks at once.\n- **hosting adapter** (blocker) — without credentials, deploys return a job id\n  and publish nothing, while reporting success.\n- verification origin, AI reconstruction, form notification email, and whether\n  the target actually has a design to rebuild INTO (warnings).\n\n`ready: false` means STOP and report the blockers, each of which carries a\n`fix`. Don't start \"and fix it after\": the content work would have to be\nredone. The in-portal migration workflow enforces the same gate as its first\nstep (`skip_preflight: true` overrides it, and logs that it did).\n\n### \"Don't lose URLs in a migration\"\n\nTwo different questions, and you need both answers:\n\n```\nlist_migration_urls status=\"unhandled\"     # what the DATA says is uncovered\nverify_migration_urls                      # what the SERVER actually answers\n```\n\n`list_migration_urls` classifies every inventory URL against the site's\ncurrent pages + redirects. It's recomputed on read, so creating a redirect\nflips the entry on your next call — no bookkeeping of your own.\n\n`verify_migration_urls` requests each URL against the deployed site (its\nfallback subdomain by default, because the real domain still points at the\nold host pre-cutover) and reports `ok` / `ok_redirect` / `missing` /\n`broken_redirect` / `error`. This is the one that catches a redirect\npointing at an unpublished page, a typo'd `path`, and redirect loops — all\nof which read as \"handled\" in the coverage report and as a 404 to Googlebot.\n**Deploy first**: it tests saved, deployed content, not your drafts.\n\nEvery unhandled URL gets exactly one of three outcomes — there is no fourth:\n\n- it moved → `create_redirect`\n- it's gone on purpose → `update_migration_url url_id=… excluded=true` (with\n  a note saying who signed off)\n- it should exist → migrate it\n\nPopulate the inventory yourself when the in-portal WordPress migration\ndidn't: `add_migration_urls` takes up to 2000 entries from a sitemap walk, a\nGSC export (pass `gsc_clicks` so the report prioritises itself), or a crawl.\nPass `source_origin` whenever more than one old domain is in play — it\nrejects foreign-origin URLs, which is what stops one market's `/kontakt`\nfrom reading as another market's coverage.\n\nFor a whole family of sites, read the `tr-migrate-multisite` skill.\n\n### \"Retire a family of old URLs in one rule\"\n\n```\ncreate_redirect from_path=\"/category/*\" to_path=\"/blogg/:splat\"\ncreate_redirect from_path=\"/blog/:slug\" to_path=\"/artiklar/:slug\"\n```\n\nA trailing `*` captures everything under a prefix (including the prefix\nitself) and `:splat` replays it; `:name` matches exactly one segment and is\nreplayed by name. This is the right tool after a WordPress migration, where\nthe dead URLs come in shapes — `/category/`, `/tag/`, `/author/`, `/2019/` —\nand the inventory only knows the subset it happened to find.\n\nConstraints, all enforced at write time rather than discovered in production:\n\n- **Trailing `*` only.** Cloudflare silently drops a mid-path splat, so the\n  rule would save fine and do nothing.\n- **`:splat` requires a `*`**, and `:name` in the target must be declared in\n  `from_path`.\n- **Query strings can't be matched** — `_redirects` keys on the path. A\n  WordPress `/?p=123` URL has to be handled at the source.\n- **A rule that would hide a live page is refused**, naming the pages.\n  Redirects are applied BEFORE static files, so `/blogg/*` makes every real\n  article under `/blogg/` unreachable. Narrow the prefix.\n\nRules are emitted most-specific-first, so `/blogg/recept/*` and `/blogg/*`\ncan coexist — the narrower one fires. `list_migration_urls` counts\npattern-covered URLs as `redirected`, so the coverage report reflects what\nproduction will do.\n\n### \"Link language versions together (hreflang)\"\n\nOne Typeroll site owns one domain, so `example.se` / `example.de` /\n`example.co.uk` are three sites. Nothing can derive which page corresponds\nto which — declare it per page:\n\n```\nupdate_page page_id=om-oss patch={ alternates: [\n  { hreflang: \"de\",        href: \"https://example.de/ueber-uns\" },\n  { hreflang: \"x-default\", href: \"https://example.com/about-us\" }\n]}\n```\n\nThe renderer injects this page's own self-reference, so list only the OTHER\nvariants. Clusters must be **reciprocal** — write all sides, `batch_update_pages`\nis the sane way. Use absolute URLs on the FINAL domains (never the\n`*.typeroll` fallback). Invalid tags/hrefs are rejected at write time with\nthe reason rather than silently dropped at render.\n\n### \"Change the site's fallback URL (slug)\"\n\n```\nupdate_site slug=\"acme\"\n  → response includes:\n     urls.fallback: \"https://acme.sites.typeroll.com\"\n     dns_note: \"New fallback URL … attached to CF Pages. SSL provisioning\n                takes 1–10 minutes after DNS propagates. …\"\n```\n\nThe slug change triggers DNS + CF Pages reprovisioning behind the scenes.\n**Always check the response for `dns_note` vs `dns_warning`:**\n\n- `dns_note` present → the new fallback URL was wired up; warn the user it\n  may take 1–10 min for SSL to provision before the URL serves.\n- `dns_warning` present → the slug was saved but DNS / CF attach failed.\n  The `urls.fallback` field is still returned (it's just `{slug}.{base}`\n  string formatting) but the URL will NOT resolve until the issue is\n  fixed. Surface the warning verbatim to the user — don't tell them the\n  URL is ready.\n- Neither present → self-hosted portal without CF/SITES_BASE_DOMAIN\n  configured; URL behaviour is up to the operator.\n\nThe old fallback URL keeps working (bookmarks + SEO survive). Customer\ncan manually deprovision the old one via the portal.\n\n## Safety boundaries\n\n- **HTML is sanitized at save.** No `<script>`, no `onclick`, no\n  `javascript:` URLs in page or partial bodies. `<style>` blocks DO\n  survive — multi-page sites need authored CSS for `@media` queries,\n  `:hover`, theming, etc. Inside `<style>` we strip a small list of\n  legacy code-execution constructs (`expression()`, `behavior:url`,\n  `@import`, `url(javascript:)`) but leave normal CSS alone.\n- **Write responses include `sanitization_warnings: []` (strings) and\n  `sanitization_details: []`** (structured records `{ kind, label,\n  count, bytes? }`). Use the structured form to programmatically retry\n  with a fixed input.\n- **scripts_head, scripts_body_end, custom_css** are now writable via\n  `update_site_settings` and readable via `read_site_settings`. Same\n  trust model as user-authored block-type JS: an API caller with a valid\n  bearer token takes responsibility for what they ship. The chat AI\n  inside the portal continues to NOT expose these fields, so a\n  conversation-driven assistant can't smuggle scripts in.\n- **The API key is site-scoped.** Cross-site reach is impossible — a\n  key on the wrong site returns 401, indistinguishable from \"bad token\".\n- **Audit log.** Every state-changing call (POST / PATCH / PUT /\n  DELETE) is logged. Reads aren't. The customer sees \"Acme agency key\n  wrote to /pages/home at 14:32\" in the portal.\n- **Rate limits.** 600 reads/min, 60 writes/min per key. On 429 the\n  response carries `Retry-After`.\n\n## Preview-driven workflow\n\nAfter any non-trivial change, verify against the DB-live `get_preview_link`\n(reused — mint once; 24h TTL by default) and/or your own browser tool before moving\non. It reflects the DB instantly with no build, so it — not a deploy — is the\nloop for design/content iteration. One reload vs. shipping a broken redesign —\nalways worth it.\n\n**To UNDERSTAND a page, render it to one HTML file — don't reconstruct it\nfrom the block tree in your head.** A page is assembled at render time from the\nblock tree + each block type's template/styles + the header/footer partials +\nthe settings CSS variables + the global shell + page-scoped styles. `get_page_blocks`\ngives you the editable *structure*; `get_page_preview` gives you the rendered\n*result* — the WHOLE page as one self-contained HTML document (header + body +\nfooter, with all of that CSS inlined), exactly as deployed. Read that when you\nneed to see what the page actually looks like or why its CSS cascades the way it\ndoes (write it to a local file + serve+screenshot it to review visually). Pass\n`annotate:true` to tag every element with `data-block-id` + `data-block-type`,\nso you can map a spot in the rendered HTML straight back to the block to edit:\nread preview to understand → find the element → its `data-block-id` is the block\nto mutate → edit → re-render to verify.\n\n**CSS precedence — where your overrides land in the cascade.** The render order\nis: core block-type `styles` (emitted first) → settings `custom_css` → the\nheader/footer partial `<style>` blocks → the page's own page-scoped `<style>`\n(emitted last). Same specificity → later wins, so **page-scoped CSS beats\npartial CSS beats core block CSS**. Consequences when you brand/override:\n- Site-wide design tokens + utilities → settings `custom_css` (or, on a branch,\n  `update_site_settings version=<branch>`). Header/footer-only tweaks → the\n  partial. One page → that page's `<style>`.\n- Core blocks set their own chrome (e.g. `core/image` gives `figure>img` a\n  `border-radius`/`margin`; `.page-content img` adds more). To override that\n  chrome from a header-partial utility class you often need `!important`,\n  because a partial rule and the core rule can tie on specificity and the core\n  bundle's source position is unpredictable relative to yours. That `!important`\n  is expected today — it is NOT a smell. (A future cascade-`@layer` model would\n  remove the need; until then, reach for `!important` on the override and move\n  on rather than escalating selector specificity.)\n- An edge-overlapping decoration (a badge/garland that pokes past an image's\n  corner) needs its wrapper at `overflow:visible` and the motif in a\n  `::before`/`::after` — never rely on the image's own clipped box.\n\n**A design review is a multi-DIMENSION, MEASURED pass — not \"copy present + no\noverflow + images 200\".** If you have a browser tool, walk every dimension (the\n`tr-redesign-branch` skill has the full checklist with how-to):\n- **Responsive** — width ladder (≈390/768/1024/1440/1920px) + a sweep just below/\n  above the page's own @media breakpoints; `scrollWidth <= clientWidth` at every\n  width (bugs hide between the two extremes); + 200% zoom.\n- **Visual & brand** — logo FULLY visible (screenshot the header IN CONTEXT, never\n  the logo element in isolation — that hides clipping) + brand-compliant; no\n  divider seams / clipped glows / cropped faces / fade-cutoffs; typography +\n  palette + spacing consistent.\n- **Accessibility (measure)** — actual contrast ratios (AA 4.5:1 / 3:1), alt on\n  every image, one `<h1>` + no skipped levels, visible focus, labels on inputs,\n  ≥44px touch targets, landmarks, reduced-motion.\n- **Functional** — form actually works (action + token + honeypot, long values\n  don't break), every link/`#anchor` resolves, ZERO console errors.\n- **Content** — no unrendered `{{…}}`, no placeholder, copy matches the live page.\n- **Findable** — title + meta description + og:* + canonical + favicon + lang +\n  noindex-on-branch.\n- **Fast** — images sized right + modern format + width/height set + lazy/eager.\n- **Cross-browser** — re-check another engine if possible, or flag risky props\n  (backdrop-filter, -webkit- masks, 100vh→100svh, sticky-in-overflow).\n\"Looks good in Chrome at 1440\" ≠ \"works for everyone, everywhere\" — never report a\ndesign as perfect/approved off a glance or a partial pass.\n\nPreview shows DB state (drafts included). Live (`get_site → urls.production`)\nshows the most recent deploy. Branch deploys live at\n`get_version → deploy_url` (`{branch}.{project}.pages.dev`).\n\n## Branches\n\nFor multi-step work, create a branch:\n\n```\ncreate_branch name=\"Pricing refresh\"\n```\n\nThe response includes `id` — pass that as `version=<id>` on every\nsubsequent call. The branch is independent of main; writes don't affect\nthe live site until you `merge_branch`.\n\nBranches default `robots_blocked: true`. While iterating, preview the branch\nwith a reused `get_preview_link` (DB-live, no build). Deploys to a branch land\nat a stable URL (`{branch}.{project}.pages.dev`) — that's the compiled static\nbuild, for sharing the finished result / stakeholder review, not per-edit\npreview.\n\n## When in doubt\n\n- **Read before you write.** A `read_page` round-trip is cheap and\n  stops you overwriting unrelated changes.\n- **Dry-run bulk operations.** `bulk_replace_text` accepts `dry_run:\n  true` and returns 3 sample diffs. Show them to the user before the\n  real run.\n- **Watch the sanitization_warnings array.** If it's non-empty, the\n  stored HTML differs from what you sent. Read it back to confirm.\n- **One small confirmation > one large undo.** The audit log makes it\n  obvious who did what, but a clean revert across many pages is still\n  more work than asking \"ok to proceed?\" first.\n- **Match the site's design.** Read a partial or two before designing\n  new components. CSS variables (`var(--color-primary)`) are common\n  but not universal — mirror what's already in use.\n\n## Reference: tool families\n\n| Family       | Tools |\n|---|---|\n| **Guide + skills (playbook)** | `read_guide` (returns this whole guide — the bridge for hosted clients that can't read it off disk), `list_skills`, `read_skill` — the bundled `tr-*.md` recipes (incl. `tr-responsive` for per-breakpoint layout). Call `list_skills` first when a task looks like \"build / migrate / redesign a site\", then `read_skill name=…`. No API key or site context needed. |\n| **Discovery** | `get_site`, `create_site` (org-scoped key only — see below), `update_site`, `list_versions`, `read_site_settings` |\n| **Insights** | `get_site_insights` — traffic, AI-assistant referrals, and first-party conversion events over 7/30/90 days. Read-only. Traffic is powered by Cloudflare Web Analytics; conversion rows come from validated Analytics attribution `click_event` targets and can be present even when the traffic provider is unavailable. |\n| **Pages — reads** | `list_pages`, `read_page`, `batch_read_pages` |\n| **Pages — writes** | `create_page`, `update_page`, `replace_page`, `batch_update_pages`, `delete_page`, `clone_page` |\n| **Pages — blocks** | `get_page_blocks`, `add_block`, `update_block`, `move_block`, `remove_block`, `set_page_mode`, `convert_page_to_blocks` |\n| **Pages — meta** | `get_page_preview` |\n| **Global blocks (partials)** | `list_partials` (summary by default), `read_partial`, `create_free_block`, `update_partial`, `replace_partial`, `delete_partial`, `find_pages_using_block`, `list_blocks_with_usage` |\n| **Block types** | `list_block_types`, `read_block_type`, `find_pages_using_block_type`, `export_block_types`, `import_block_types` |\n| **Collections** | `create_collection`, `update_collection_schema`, `delete_collection`, `list_collections`, `read_collection`, `list_collection_items` (richtext hidden by default), `read_collection_item`, `batch_read_collection_items`, `create_collection_item`, `update_collection_item`, `delete_collection_item`, `regenerate_collection_listing` |\n| **Media** | `list_media`, `read_media`, `create_upload_url`, `upload_media_from_url`, `upload_media_inline`, `update_media`, `delete_media`, `finalize_media`, `finalize_all_media`, `generate_image_variants`, `suggest_alt_text_context` |\n| **Redirects** | `list_redirects`, `create_redirect`, `delete_redirect`. `from_path` may be a PATTERN: a trailing `*` (with `:splat` in the target) or `:name` for one segment — one rule retires a whole family of dead URLs (`/category/*` → `/blogg/:splat`). Mid-path splats and query strings are refused, as is any rule that would hide a live page. |\n| **Migration inventory** | `get_migration_readiness` (preflight — CALL FIRST), `list_migration_urls`, `add_migration_urls`, `update_migration_url`, `update_migration_urls`, `delete_migration_url`, `import_sitemap`, `import_gsc_performance`, `verify_migration_urls`. Sitemap indexes are recursive. GSC supports direct Search Console access or CSV and aggregates fragment variants. Verification is compact by default. |\n| **Forms** | `list_forms`, `read_form`, `create_form`, `update_form`, `delete_form`, `list_form_submissions`, `delete_form_submission` (removes one submission — e.g. cleaning up a test entry; `delete_form` with `delete_submissions` is the bulk path). **Steps (form/* block trees) are the ONLY stored model**: pass `steps` for funnels, or `fields` for simple forms — the server converts a flat field list to a single static step. Place with a `core/form` block on block-mode pages or `<x-form id=\"…\" />` in HTML mode. Both expand server-side to the same complete signed shell and initial state. Email/webhook actions are admin-only in the portal and excluded from agent reads/writes. |\n| **Settings** | `update_site_settings` (whitelist) |\n| **Core modules** | `list_apps`, `read_app`, `update_app` (legacy API name; admin; schema-driven config, masked secrets, redeploy when `affects_build` is true) |\n| **Extension installations** | `list_extension_installations`, `read_extension_installation`, `update_extension_installation_config` (admin; schema-driven config, masked secrets preserved, redeploy after frontend-facing changes) |\n| **Analytics attribution** | `read_funnel_attribution`, `update_funnel_attribution` (specialized Analytics module tools; admin; redeploy after changes) |\n| **Search + bulk** | `search_pages`, `check_internal_links`, `bulk_replace_text`. The link check is database-driven. Bulk replace defaults to pages but can target partials, collection items or all resources, always dry-run first. |\n| **Branches** | `create_branch`, `read_version`, `delete_branch`, `merge_branch` |\n| **Deploy** | `trigger_deploy`, `list_deploys`, `get_deploy_status` |\n| **Preview** | `get_preview_link`, `get_page_preview` |\n\nEvery tool's input is validated server-side; the MCP server only does\nauth + shape. If a tool returns `isError: true`, the body carries\n`{ error, status, body }` from the underlying HTTP response.\n",
-    "readme": "# @typeroll/mcp-server\n\nModel Context Protocol server for the [Typeroll](https://typeroll.com)\npublic API. Lets Claude (Desktop / claude.ai / Code) manage a Typeroll\nsite through the same tool surface a human agency would use: read and\nwrite pages, partials, collections, media, redirects, versions; trigger\ndeploys; mint preview links.\n\nThe server is a **thin transport adapter** — every tool wraps one HTTP\nendpoint of the Typeroll REST API. Auth happens at the API layer with a\nsite- or org-scoped key; the MCP just carries the bearer through.\n\n## Two ways to connect\n\n- **Hosted (Claude Desktop / claude.ai) — paste a URL.** No CLI, no\n  Node.js install. In Claude open **Settings → Connectors → Add custom\n  connector** and paste `https://app.typeroll.com/api/mcp`\n  (or `https://<your-self-hosted-portal>/mcp`). Claude opens a consent\n  page; paste your Typeroll API key there.\n- **Stdio (Claude Code) — one `claude mcp add` command.** Best for local\n  dev / agency staff already in a terminal. Instructions below.\n\nThis npm package is the stdio transport. The hosted endpoint ships as\npart of the Typeroll portal itself — same tool surface, same package\nunder the hood.\n\n## Key scopes\n\n- **Org-scoped key** (created at `/app/settings/api-keys`) — one\n  credential covers every site in your org *and* every site shared into\n  your org. The default for the hosted Claude connector. Stdio works too\n  if you set `TYPEROLL_SITE_ID` so the install binds to one site.\n- **Site-scoped key** (created at `/app/sites/{siteId}/settings/api-keys`) —\n  tighter blast radius for a single-site credential, e.g. one you'd\n  hand to a customer for a self-managed site.\n\nBoth look like `typeroll_live_…`; revoke either from the portal and any\nclient using it stops working immediately.\n\n## Stdio quick start (Claude Code)\n\n1. **Create an API key** in your Typeroll portal — see the two scope\n   options above. Org-scoped is the right default.\n\n2. **Add the server to Claude Code.** Drop this into `~/.claude.json`\n   (or your local `.claude/config.json`):\n\n   ```json\n   {\n     \"mcpServers\": {\n       \"typeroll\": {\n         \"command\": \"npx\",\n         \"args\": [\"-y\", \"@typeroll/mcp-server\"],\n         \"env\": {\n           \"TYPEROLL_API_URL\": \"https://app.typeroll.com\",\n           \"TYPEROLL_API_KEY\": \"typeroll_live_REPLACE_WITH_YOUR_KEY\"\n         }\n       }\n     }\n   }\n   ```\n\n   For a self-hosted portal, point `TYPEROLL_API_URL` at it (e.g.\n   `https://cms.example.com`).\n\n   Prefer a scaffold? Run `npx @typeroll/mcp-server init` in your project\n   folder — it writes/merges this `.mcp.json`, copies the skills into\n   `.claude/skills/`, and adds an `AGENTS.md` pointer + imagegen-lab\n   files. Idempotent; `--force` to overwrite. (Skills only:\n   `npx @typeroll/mcp-server install-skills .claude/skills`.)\n\n3. **Tell the agent what kind of work you want.** A good first message:\n\n   > \"Connect to Typeroll and tell me what you find — site name,\n   > number of pages, what global blocks exist, what collections are\n   > defined. Then I'll give you a task.\"\n\n   Claude will call `get_site`, `get_site_capabilities`, `list_pages`,\n   `list_partials`, `list_collections`, and `list_block_types` in sequence and\n   report back. The capabilities + block palette are mandatory before it\n   chooses HTML mode or reports a missing site-building feature.\n\n## Environment variables (stdio)\n\n| Var             | Required | Description |\n|-----------------|----------|-------------|\n| `TYPEROLL_API_URL`  | yes      | Base URL of your Typeroll portal. |\n| `TYPEROLL_API_KEY`  | yes      | A `typeroll_live_…` bearer token. |\n| `TYPEROLL_SITE_ID`  | sometimes | Pin to a specific site. Required when using an org-scoped key over stdio (the install can only target one site at a time); auto-detected for site-scoped keys. |\n\n## Extension developer CLI\n\nThe package also installs `typeroll`. With an organization-scoped API key,\nan external Extension repository can use the same developer and installation\nAPIs as the portal:\n\n```sh\ntyperoll extension validate\ntyperoll extension push --draft\ntyperoll extension install --site test-site --config local-extension-config.json\ntyperoll extension promote 1.0.0\n```\n\nThe manifest defaults to `typeroll-extension.json`; use `--manifest` to select\nanother file. Local validation is a fast preflight. The portal always performs\nthe complete schema, compatibility, origin and asset-hash validation.\n\n## What the agent should read first\n\nThe package ships [AGENTS.md](./AGENTS.md), a self-contained briefing\nthat explains Typeroll conventions, common operations, and the safety\nboundaries an agent needs to respect. Point Claude at it (or include it\nin your project's `CLAUDE.md` / `AGENTS.md`) so it knows when to use\nwhich tool.\n\n## Tool surface\n\nMore than 100 tools across these families. See [AGENTS.md](./AGENTS.md) for\nthe full reference + concrete operation recipes.\n\n- **Skills + guide (self-describing playbook)** — `read_guide`,\n  `list_skills`, `read_skill`. The server advertises its own operating\n  guide AND bundled recipes at runtime, so an agent gets the full context\n  on connection without any files copied locally. `read_guide` returns the\n  whole AGENTS.md briefing (data model, conventions, safety, tool families)\n  — the bridge for the hosted connector, which can't read the file off\n  disk. `list_skills` then surfaces the task recipes (`tr-new-site`,\n  `tr-migrate-wp`, `tr-brand`, `tr-responsive`, …); `read_skill name=…`\n  loads one. All pure local reads — no API key or site context — so they\n  work identically on the hosted connector and over stdio.\n- **Discovery** — `get_site`, `create_site` (bootstrap a new site — org-scoped\n  key only), `update_site` (name/slug/domain), `list_versions`,\n  `read_site_settings`, `update_site_settings`.\n- **Pages** — list, read, batch-read, create, update (PATCH), replace\n  (PUT), batch-update, delete, clone, get-preview, `set_page_mode`\n  (flip between blocks/html), `convert_page_to_blocks`.\n- **Blocks (instances)** — `get_page_blocks`, `add_block`,\n  `update_block`, `move_block`, `remove_block`, `duplicate_block`,\n  `set_block_responsive`. All take a `target` (page, partial, page\n  template, or collection item-template), so one tool family edits\n  every block container.\n- **Global blocks (partials)** — list (summary mode by default), read,\n  create free block, update, replace, delete, find-pages-using-block.\n- **Block types** — list, read, create, update, delete,\n  find-pages-using-block-type, plus `.tcblocks` export/import. Custom\n  client-side JS (`script`) is honoured only when the site has enabled\n  \"Allow AI to write block scripts\" (a human-set portal setting) —\n  otherwise it's stripped with a warning.\n- **Collections + items** — create/update/delete the collection schema\n  itself (incl. `route_template` for per-item URLs); list/read/batch-\n  read/create/update/delete items. Existing-collection tools consistently\n  use `collection` (the old `name` argument remains accepted as an alias).\n  Collection repeaters support `group_by`, and item templates can place\n  `template/item_navigation` for deterministic previous/next links.\n- **Media** — list, read, signed upload URLs, `upload_media_from_url`,\n  `upload_media_batch_from_urls` (1–50 sources, max 25 MiB each, with\n  partial-success results),\n  `upload_media_inline` (all URL uploads auto-finalize after PUT — see below),\n  patch metadata, delete, `finalize_media` (per-item: applies immutable\n  Cache-Control + generates AVIF/WebP srcset variants — call after\n  `create_upload_url`'s raw PUT path), `finalize_all_media` (bulk\n  backfill for legacy libraries), `generate_image_variants` (the\n  variant half of finalize, kept for surgical reruns),\n  `suggest_alt_text_context` (returns a tuned prompt for your own\n  vision model).\n- **Rendering controls** — `core/table_of_contents`, per-site\n  `trailing_slash`, exact `iframe_allowed_hosts`, `icon_192`, and per-page\n  `append_seo_suffix=false`. The block editor supports labelled enums,\n  line-based lists, nested repeating arrays and an internal-page URL picker.\n- **Redirects** — list, create, delete. Plus automatic 301 on slug change.\n- **Forms** — list, read, create, update, delete, list submissions.\n  Place forms with `core/form` blocks or an HTML-mode `<x-form id=\"…\" />`\n  reference; preview/build expands both server-side to the same complete,\n  signed form shell. Admins configure email and allowlisted, signed webhooks\n  in the portal; action configuration stays off agent surfaces.\n- **Extension installations** — list/read installed Extensions and update\n  manifest-defined installation config through the API key with\n  `update_extension_installation_config`; omitted and masked secrets are\n  preserved. Use this for frontend config such as consent copy and policy\n  links, then redeploy.\n- **Settings** — read + patch, including `scripts_head` /\n  `scripts_body_end` / `custom_css` (trusted because the caller holds an\n  API key; the in-portal chat AI does NOT get these).\n- **Core modules** — list the legacy `apps` registry, read schema + masked\n  state, and enable, configure, or disable any module with the same admin API key used for content\n  and deploys. Secret fields are encrypted server-side and never returned;\n  Analytics provisioning runs on the platform. Deploy after updates whose\n  response has `affects_build: true`.\n- **Search + link integrity** — `search_pages` plus `check_internal_links`,\n  which resolves saved database content against pages, collection/facet routes,\n  media and redirect chains without crawling the public site.\n- **Bulk** — `bulk_replace_text` with dry-run across pages, partials,\n  block data and schema-defined collection-item fields.\n- **Migration inventory** — bulk add/update decisions, recursive\n  `import_sitemap`, direct or CSV-fallback `import_gsc_performance`, and compact\n  `verify_migration_urls` (successful rows omitted unless requested).\n- **Branches** — create, read, delete, merge. Branch deploys get their\n  own URL at `{branch}.{project}.pages.dev`.\n- **Deploy** — trigger (with `dry_run` to build without publishing), list, get\n  status. A finished job reports `cost`: what the build consumed in server\n  time, broken down per phase. Estimates from a rate card, not billing records.\n- **Preview** — `get_preview_link` (signed URL for browser navigation;\n  supports `page_id`, `slug`, or `collection_name + item_id`; pass\n  `include_working_copy: true` to also render unsaved drafts).\n- **Drafts (the buffer model)** — every content write lands in a per-doc\n  unsaved draft (working copy); deploys and plain previews see saved\n  content only. Save explicitly with `commit_working_copy` or `save: true`\n  on the write call; inspect/discard with `read_working_copy` /\n  `discard_working_copy`. Status changes and structural operations apply\n  immediately.\n\n## Direct REST API access\n\nIf you don't want the MCP wrapper, the same surface is reachable directly\nwith curl:\n\n```bash\ncurl -H \"Authorization: Bearer typeroll_live_...\" \\\n  https://app.typeroll.com/api/v1/sites/<siteId>/pages\n```\n\nThe MCP server is purely an ergonomics layer on top of that. The complete v1\ncontract, including payload envelopes and collection-item slug addressing, is\ndocumented in [`docs/v1-api.md`](../../docs/v1-api.md).\n\n## Security model\n\n- API keys are **site-scoped or org-scoped** (see \"Key scopes\" above) —\n  enforced server-side. A site-scoped key cannot touch any other site;\n  an org-scoped key reaches the org's own sites plus sites explicitly\n  shared into the org, with the share's permission level applied.\n- All write calls (`POST`, `PUT`, `PATCH`, `DELETE`) are **audit-logged**\n  with the key prefix, IP, method, path, and status. Reads are not\n  logged (cost vs. value).\n- **Rate limits**: 600 reads/min, 60 writes/min per key. 429 responses\n  carry `Retry-After` headers.\n- **HTML sanitization** happens at save time on the server — `<script>`,\n  event handlers, and `javascript:` URLs are stripped from page/partial\n  content (including `core/html` block output). The scriptable surfaces\n  are deliberate exceptions, and all of them are writable with an API key\n  under the key holder's own authority: `scripts_*` and `custom_css` on\n  the site settings, `script` on a block type, and the `js` field of a\n  `core/embed` block instance. Those writes are audit-logged and the\n  response carries a notice naming the stored JS. Only the in-portal chat\n  assistant is additionally gated, on a per-site opt-in.\n- Keys can be **revoked** at any time from the portal. Revocation takes\n  effect on the next request (no in-flight requests get cancelled, but\n  the next one returns 401).\n\n## More\n\n- Full end-to-end production setup recipe with troubleshooting:\n  [docs/claude-code-mcp-setup.md](../../docs/claude-code-mcp-setup.md)\n- Agent operations briefing: [AGENTS.md](./AGENTS.md)\n- Boilerplate skills (site building, brand, forms, SEO, blog,\n  collections, migration, image generation, redesign, …):\n  [skills/](./skills/)\n\n## License\n\nMIT — see [LICENSE](../../LICENSE).\n"
+    "agents": "# AGENTS.md — Working on a Typeroll site\n\nYou are connected to a Typeroll site through `@typeroll/mcp-server`.\nThis file is your briefing: what the system is, what conventions matter,\nwhat tools to reach for first.\n\nIf anything below conflicts with what you observe in the tools, trust the\ntools — the platform may have moved since this was written.\n\n**Start here for site-shaped tasks.** When the user wants to build,\nmigrate, redesign, or brand a site, call `list_skills` first — the server\nadvertises its own step-by-step playbook (`tr-new-site`, `tr-migrate-wp`,\n`tr-brand`, …). Then `read_skill name=…` loads the full recipe. These are\nlocal reads; no API key or site context required.\n\n**Branch first for anything larger than a small edit.** Before a redesign,\na multi-page change, or trying out a new design direction, run\n`create_branch name=\"…\"` and pass the returned id as `version=<id>` on every\nsubsequent read/write. The work stays off the live `main` version until you\n`merge_branch` it — nothing ships until you decide it should. Branches default\n`robots_blocked:true` and get their own deploy URL for stakeholder review.\nIt's the cheapest insurance there is; when in doubt, branch. The\n`tr-redesign-branch` skill walks the whole flow. (Small, low-risk single edits\ncan go straight to main.)\n\n## What this is\n\nTyperoll is a static-site CMS: content lives in a database, the user\nedits it through an in-app editor, and a deploy step compiles everything\nto a fast static site hosted on Cloudflare Pages. The in-app chat handles\nsingle-page or single-block edits by the editor audience. You — through\nthis MCP — handle the work that doesn't fit there: site-wide redesigns,\nbulk content updates, structural migrations, directory imports.\n\nThe MCP server is a thin wrapper around the public REST API. Each tool\nmaps to one HTTP endpoint; the actual logic runs in the customer's portal\n(SaaS or self-hosted).\n\n## The data model in 90 seconds\n\n- **Pages.** Title, slug, status (`draft | review | unlisted | published`),\n  body content + SEO fields. Two body shapes selectable per page via\n  `content_mode`:\n  - `blocks` (DEFAULT for new pages) — `blocks: Block[]` tree of typed\n    blocks (heading, prose, section, columns, image, button, plus any\n    user/third-party block types installed on the site). Use the\n    block-mutation tools (`add_block`, `update_block`, `move_block`,\n    `remove_block`) for structural changes.\n  - `html` — body lives in `html_content` as a single HTML string.\n    Useful when you have hand-written markup to drop in directly.\n\n  Slug is a single path segment — no slashes. `about` → `/about`,\n  `kontakt` → `/kontakt`, empty string `\"\"` → homepage. The v1 API\n  rejects `services/design` and other slash-containing slugs with\n  \"Invalid slug … slugs must not contain slashes.\" For nested URLs\n  like `/blog/{slug}` or `/services/{slug}`, the right primitive is a\n  **collection with `route_template`** (see the `tr-blog` and\n  `tr-directory` skills) — not a flat page with a slashed slug.\n\n- **Partials = global blocks.** Three kinds:\n  - `header` — auto-injected at the top of every page.\n  - `footer` — auto-injected at the bottom of every page.\n  - `free` — reusable HTML you drop into a page with\n    `<x-include name=\"block-id\" />`. Free blocks are how you avoid\n    duplicating HTML across HTML-mode pages.\n\n  Partials themselves also support `content_mode='blocks'` — pass a\n  `blocks: Block[]` tree to `update_partial` and the renderer composes\n  it the same way as a page. Useful for header/footer authored with\n  block types.\n\n- **Collections.** Repeatable content types (blog, team, events,\n  products, restaurants for a directory site, etc.). Each has a schema\n  (`fields[]`) and optional **per-item routing** via `route_template`\n  (e.g. `/restaurants/{slug}`). When set, every published item gets its\n  own static URL rendered through `item_template_html`. Set\n  `route_template=\"\"` to opt out and keep the collection listing-only.\n\n- **Settings.** Site name, tagline, logo, favicon, colors, fonts,\n  contact info, social links, SEO suffix, default meta description\n  (`default_meta_description` — site-wide fallback for pages without a\n  `seo_description`; tagline is the last resort), plus `scripts_head`,\n  `scripts_body_end`, `custom_css` (writable via the API — your bearer\n  token authorises shipping arbitrary CSS/JS to the live site, just\n  like editing a partial's HTML does).\n\n- **Analytics attribution.** `read_funnel_attribution` and\n  `update_funnel_attribution` manage the Analytics module's validated,\n  site-level rules that forward\n  allowlisted query parameters to exact HTTPS link targets. Optional\n  first-/last-touch storage is consent-gated and off unless configured. Read\n  the current config before writing. For advertising pass-through, preserve\n  only incoming or stored values and omit `fallback`: a fallback creates\n  synthetic attribution when no campaign value exists. The API rejects such\n  fallbacks unless `allow_synthetic_fallbacks=true` explicitly acknowledges\n  that behavior. A target `click_event` is recorded as a consent-gated,\n  first-party conversion when Analytics is enabled; the event contains only\n  the declared funnel, destination, page path, and resolved allowlisted\n  attribution. Navigation never waits for analytics.\n\n- **Core modules.** `list_apps`, `read_app`, and `update_app` expose the\n  code-defined core-module registry (the `apps` API name is retained for\n  compatibility) through the same admin API key used for content\n  and deploys. Read the schema before writing. Secret fields stay masked on\n  reads and encrypted at rest; omitted fields preserve their current values.\n  When `affects_build` is true, deploy after the update.\n\n- **Extension installations.** `list_extension_installations`,\n  `read_extension_installation`, and `update_extension_installation_config`\n  expose each installed Extension's manifest-defined config through the admin\n  API key. Read the installation before writing so you use the exact keys from\n  `manifest.config_schema`. This is the supported automation path for public\n  content such as consent copy, link text, and policy URLs; masked secrets are\n  preserved when omitted. The update queues a production deploy by default;\n  pass `deploy: false` only when batching several changes and deploy once after\n  the final update.\n\n- **Page templates.** A `PageTemplate` is a Block[] tree that wraps a\n  page's body. The template contains exactly one block of type\n  `template_content_slot` — at render time that block gets replaced by\n  the page's own `blocks`. Set `Page.template = \"<template-id>\"` to\n  apply a template to a page.\n\n- **Block types.** A site has three sources of block types:\n  - **Core** (origin: 'core', ids like `core/section`) — shipped in\n    the platform, always available.\n  - **User** (origin: 'user') — created in the portal's block-types UI.\n  - **Third-party** (origin: 'third_party') — imported from .tcblocks\n    packages via `import_block_types`.\n\n  `list_block_types` returns ALL of them in one list as a lightweight\n  summary: each entry's id, label, category, container/slot info, origin,\n  and full field schema (names, types, defaults) — but NOT the render-time\n  template/styles/script (omitted so the list stays within token budget as\n  the library grows). Use `read_block_type` for one block's markup, or pass\n  `full:true` to inline it for every block. Always call this FIRST before\n  working with blocks — never hardcode block ids or field names, the\n  available set is per-site.\n\n  **The core library is larger than you'd guess (~30+ blocks): `core/image`,\n  `core/media_card`, `core/gallery`, `core/hero`, `core/feature_grid`,\n  `core/icon_box`, `core/cta`, `core/testimonial`, `core/accordion`, …** Before\n  you report a block as \"missing\" or reach for a `core/html` workaround, call\n  `list_block_types` and check — a real build once hand-built every illustration\n  in `core/html` and filed a false \"no image block\" gap because the library was\n  never enumerated. Prefer a native block; `core/html` is the last resort.\n\n  Block-library specifics worth knowing (template_capabilities_version\n  0.15.0):\n  - **`core/media_card`** — image + text side by side (image left/right,\n    width third/two-fifths/half, heading + richtext + button, optional\n    card background/radius; stacks image-on-top below 720px). Use it for\n    the classic \"photo next to copy\" layout instead of hand-building\n    section+grid+html.\n  - **`core/search`** (0.29.0+) — site search over the deployed site.\n    Place the block anywhere; the deploy pipeline detects it, runs\n    Pagefind over the built HTML, and the block loads the search UI at\n    visit time. Index = page content only (nav/footer excluded); noindex\n    pages stay out. Editor preview shows a placeholder note (the index\n    only exists on the deployed site).\n  - **Archive pagination** (0.29.0+): a collection listing\n    (`core/collection_list` / `core/repeater`) with `paginate: N` renders\n    N items per page + a pager, and the build generates `/page/2/`… routes\n    automatically. `paginate` supersedes `limit`; one paginated listing\n    per page.\n  - **`core/feature_row`** (0.29.0+) — the full-width \"zig-zag\"\n    feature/step row: balanced halves that hug the center gutter (no\n    wide-screen dead air), natural-aspect image (never cover-cropped —\n    that's media_card's card look), eyebrow + heading + richtext +\n    button pair, `image_side: left|right` per row, `stack_order`\n    controls what comes first on mobile. Prefer it over `core/columns`\n    with an unbalanced ratio + width-capped text for these rows.\n  - **`core/hero` and `core/cta` render their buttons server-side** via\n    `primary_label`/`primary_url` + `secondary_label`/`secondary_url`.\n    (The old `buttons` array relied on client hydration that never\n    existed — if you see `data-buttons` in stored content it renders\n    nothing; rebuild with the explicit fields.)\n  - **`core/image` gets responsive `<picture>` automatically at build\n    time** — the deploy pipeline's SEO transform converts CDN `<img>`\n    into `<picture>` with AVIF/WebP srcset variants. You do NOT need\n    `core/html` for responsive images; just point `src` at an uploaded\n    media URL (run `generate_image_variants` first) and optionally set\n    `radius`. Note: the in-portal preview shows the plain `<img>` — the\n    `<picture>` upgrade appears on the deployed site.\n  - **Icons render inline SVG** (since template_capabilities_version\n    0.16.0). Every `type: 'icon'` schema field — on `core/icon`,\n    `core/icon_box`, `core/step_card`, and custom block types — renders\n    a stroke-based inline SVG when the value is a name from\n    `get_site_capabilities → core_icon_names` (a curated Lucide subset:\n    `check`, `star`, `shield-check`, `mail`, `arrow-right`, `zap`,\n    `truck`, `chart-line`, …). Any other value (emoji, plain text) is\n    rendered as escaped text, so emoji stand-ins keep working. Icons\n    size with `font-size` (the SVG is 1em) and paint with\n    `currentColor`. Custom block templates opt in by placing the derived\n    raw token `{{{<field>_svg}}}` where the icon should appear. On\n    pre-0.16.0 portals icons don't render — use emoji or CSS markers.\n    `core/tabs` label icons are the remaining gap (tab strip is built\n    client-side).\n  - **Grids with a partial last row: set `last_row: 'center'`** (since\n    template_capabilities_version 0.16.5). Five equal cards in a 3-col\n    `core/grid` (or 7 in 4, ...) left-align the orphans by default; with\n    `last_row: 'center'` the last row auto-centers. THE DESIGN RULE: when\n    N peer cards don't divide by the column count, center the last row or\n    change the column count — NEVER invent a \"wide\"/full-width variant of\n    one peer card just to fill the hole. Special treatment is a content\n    decision, not a layout patch.\n  - **`core/section` is natively full-bleed on block pages** (since\n    template_capabilities_version 0.14.0): the section's background runs\n    edge-to-edge and meets the header with zero gap; content inside is\n    constrained by the section's own inner container (`width` field:\n    narrow/normal/wide/full). Never use 100vw negative-margin hacks.\n    Top-level blocks that are NOT sections still get a classic centered\n    container as fallback. Anchor ids and custom classes via\n    `style_overrides` are safe on full-bleed sections since 0.15.3 —\n    they merge into the `<section>` element itself. On 0.14.x–0.15.2\n    they wrapped the section in a `<div>`, which silently disabled\n    full-bleed for that section.\n  - **Shaped section transitions** (since template_capabilities_version\n    0.24.0): `core/section` takes `divider_top` / `divider_bottom`\n    (`none | wave | curve | tilt`). The platform paints the divider in the\n    section's OWN `background` and overlaps the neighbour by 1px, so a\n    cream↔colour transition renders seam-free. **Use this for waves/curves —\n    never hand-roll a divider band in `core/html`** (a separate stacked shape\n    seams against the next section as a sub-pixel hairline in Chrome). Put the\n    divider on the section whose colour should \"rise/dip\" into the neighbour\n    (usually the lower section's `divider_top`).\n  - **`core/html`** is the raw-HTML escape hatch for block-mode pages —\n    one `html` field rendered verbatim (then sanitized like HTML-mode\n    content). Use it for the genuinely unique thing no block covers.\n    Prefer real blocks when one fits.\n  - **Structured records at scale** (template_capabilities_version ≥ 0.31.0).\n    Four things landed together for directory-shaped sites:\n    - `collection_completeness` — **start an enrichment pass here**, not by\n      paging every item. Returns per-field gap counts plus the N worst\n      records (missing fields, never-verified fields, fields whose last write\n      is older than the staleness window), computed at read time. Fields no\n      API key may write are excluded by default: a gap you can't close is\n      noise.\n    - **Per-field write authority.** A collection field can declare\n      `writable_by` (`portal | owner | agent | app | import`). A write you're\n      not permitted, or one that would overwrite a higher-precedence writer\n      (a human correction, the listed business's own edit), comes back as\n      **409 with the losing field names** — never a silent no-op. Treat that\n      as \"already handled\" and record it; retrying will lose again.\n    - **Item references.** `item_ref` / `item_ref_list` fields point at items\n      in another collection (`ref_collection`). The reverse direction is\n      computed at render time — don't try to maintain backlinks yourself.\n      Render them with a `core/repeater` whose `source_type` is `related`\n      (a ref field on the current item) or `backlinks` (who points at it).\n    - **Taxonomy pages.** `CollectionDef.facets` generates one page per\n      distinct field value. ⚠️ This turns record count into ROUTE count, and\n      route count is what the build timeout measures. `min_items` (default 2)\n      keeps thin-content pages out, and combination pages must be listed\n      explicitly in `facet_combinations` — never assume a cartesian product.\n  - **`core/embed`** (template_capabilities_version ≥ 0.30.0) is\n    `core/html` plus behaviour: an `html` field and a `js` field. Reach\n    for it when a one-off placement needs JavaScript. **A `<script>` tag\n    written into `core/html` — or into any page/block markup — is stripped\n    by the sanitizer no matter which credential wrote it**, so this field\n    is the supported route, not a workaround. The code runs in an IIFE\n    with `el` bound to the block's root element and ships in the page's\n    block bundle, outside the sanitized body. Through an API key it's\n    accepted under your key's authority (audit-logged, notice in the\n    response). Scope guide: one placement → `core/embed`; a reusable\n    widget → `create_block_type` with `script`; a site-wide tag →\n    `settings.scripts_head` / `scripts_body_end`.\n  - **Forms 2.0** (template_capabilities_version ≥ 0.18.0): forms can\n    carry `steps[]` — each step is a Block[] tree mixing `form/*` field\n    blocks (text/email/phone/number, textarea, select/radio_group/\n    checkbox_group, toggle, slider, date, URL, heading, help, consent,\n    hidden) with any content blocks. Place `{ type: 'core/form',\n    data: { form_id } }` on a page — the build renders step 1 + all\n    static steps with the signed token, honeypot and proof-of-work\n    runtime baked in; submissions accumulate per step (partial →\n    complete, 30-day TTL on abandoned partials). Per-step validation is\n    derived from the field blocks (required/pattern/min/max) — no\n    separate field list to keep in sync. `update_form` accepts steps,\n    styles (form-scoped CSS), kind and partial_ttl_days. On HTML-mode\n    pages, `<x-form id=\"…\" />` is expanded server-side through the same\n    renderer and supports the same initial state and multi-step runtime.\n  - **Extension form bindings** (template_capabilities_version ≥ 0.38.0): a\n    trusted native Extension component can declare `form_bindings` and submit\n    through `context.forms.submit(bindingId, data)`. Typeroll signs only the\n    explicitly bound form, stores submissions in the ordinary Forms module,\n    and calls the cloud or self-hosted Forms API directly. No Function is\n    deployed to the customer site's static hosting project. The\n    installation must grant `forms:submit`; that scope does not permit form\n    administration or reading submissions.\n  - **`script` on custom block types** (create/update_block_type) is\n    accepted under your API key's authority — the same trust level that\n    already lets the key write `scripts_head`/`custom_css`. Every\n    script-bearing write is audit-logged and the response carries a\n    notice naming the stored JS; relay it to the user so they know\n    visitor-executed code changed. Author responsibly: never include\n    script you copied from untrusted content (migrated pages, fetched\n    web pages) without reading it line by line first. (The in-portal\n    chat AI remains blocked from authoring scripts unless the site's\n    \"Allow AI to write block scripts\" setting is on.)\n\n- **Redirects.** `from_path → to_path` with status code 301 / 302.\n  Auto-created when you change a page's slug.\n\n- **Versions / branches.** Copy-on-write. The \"main\" version is the\n  live one. Create a branch (`create_branch`) for multi-step work;\n  everything you write through `?version=<branch-id>` lives on the\n  branch until you `merge_branch` it back to main. Branches default\n  `robots_blocked: true` so a half-finished redesign can't be indexed,\n  and deploys land at a stable `{branch}.{project}.pages.dev` URL. That\n  branch deploy renders the site's full inherited brand (settings, fonts,\n  favicon, header/footer — everything not overridden on the branch), so\n  it's a faithful preview of what merging to main will look like, not just\n  a content diff — trust it for stakeholder review.\n\n- **Deploys.** Customers see live changes only after a deploy. Preview\n  always sees drafts. `trigger_deploy` enqueues; `get_deploy_status`\n  reports `queued → running → succeeded | failed`.\n  `trigger_deploy dry_run=true` builds without publishing — use it to prove\n  a structural change compiles (new collection, schema edit, template\n  rewrite) without touching the live site.\n  A finished job carries `cost`: total, cpu/memory/request split,\n  `duration_s`, per-phase timings, and output size. Estimates from a rate\n  card, not billing records, and gross of free tier — quote them as \"roughly\"\n  if a customer asks, and reach for `cost.phases` when the question is *why*\n  a build got slow.\n\n- **Is the site live?** There is no site-level status field —\n  `Site.status` was removed in 0.30.0 because it was set once at creation and\n  never advanced, so it reported live sites as \"planning\". Read\n  `get_site → urls.production` instead: non-null means the domain is verified\n  and serving. For \"has anything shipped\", use `list_deploys`.\n\n- **Site URLs.** `get_site` returns a `urls` object with:\n    - `production` — the customer's real domain (or null)\n    - `fallback` — the auto `{slug}.typeroll.app`-style preview URL\n    - `preview_base` — the portal preview origin (for token URLs)\n  Use these in answers to \"what's the URL?\" — never invent.\n\n- **For design/content iteration, share the DB-LIVE preview — don't deploy.**\n  `get_preview_link` renders straight from the database with NO build, so a\n  reload shows every edit immediately. Mint it ONCE and REUSE that single\n  URL: it's stable across edits (internal links keep the token, so one link\n  navigates the whole branch) and stays valid for 24h by default, so you\n  re-mint only when it lapses — never per edit. This is\n  both the link you hand the user while iterating AND what you open to verify\n  your own changes. Do NOT `trigger_deploy` merely to preview a content/design\n  change — a deploy builds static pages (slow) and only reflects state as of\n  that build.\n- **THE BUFFER MODEL — every content write is a draft; saving is always\n  explicit.** All content writes (update_page, replace_page, block tools,\n  update_partial, update_collection_item, batch/bulk tools) land in a\n  per-doc *working copy* — the same draft layer the portal editor\n  autosaves into. Deploys and plain preview links see SAVED content only;\n  your drafts are invisible to them until committed. The loop:\n    1. Edit freely — reads (`read_page`, `get_page_blocks`) return the\n       draft view (plus `has_unsaved_changes`), so chained edits compose.\n    2. Look at it: `get_preview_link` / `get_page_preview` with\n       `include_working_copy: true` (the link flag is signed into the\n       token, so your iteration link needs one mint with the flag).\n    3. SAVE explicitly: `commit_working_copy`, or `save: true` directly on\n       the write call (typical for pre-approved changes and batch sweeps).\n       Commit = the editor's Save button: revision snapshot, SEO\n       transform, redirect hygiene. Rejected → `discard_working_copy`.\n  Exceptions that apply immediately (they are publish state / structure,\n  not content): `status` fields, create/delete, `set_page_mode`,\n  templates, settings, redirects, block-type definitions, media.\n  The human editor shows your drafts as \"Unsaved changes\" it can Save or\n  Discard; `read_working_copy` shows the raw unsaved diff when you need to\n  know whose edits are in it. Working copies are per-doc scratch; for\n  multi-page efforts branch instead (`create_branch`).\n  **Before `trigger_deploy`: commit.** Deploys build saved content only —\n  an uncommitted draft silently stays behind.\n- **Deploys / `{branch}.{project}.pages.dev` are the STATIC BUILD**, refreshed\n  only by `trigger_deploy`. Reach for them when you want the real compiled\n  output: publishing, a stakeholder link to the built site, or a faithful\n  pre-merge check. The branch alias is permanent across re-deploys; the\n  per-deploy `{hash}.pages.dev` is immutable per build. Reserve deploys for\n  these — not for previewing edits.\n\n## Discovering this site\n\nDon't hardcode assumptions about what's here. Every fact about the site\ngoes through the MCP:\n\n**Capability discovery is a required gate for every build, migration and\nredesign.** Before choosing HTML mode, hand-writing a component, or reporting\nthat Typeroll lacks a feature, call `get_site_capabilities` and\n`list_block_types` (`full:true` only when you need every template). If a likely\ntype appears, call `read_block_type` and inspect its schema. A capability gap is\nvalid only after those reads show that neither a core/site block nor a\ncomposition of `core/section`, layout blocks, `core/repeater`, template blocks,\nor a custom block type can express the requirement. Record the calls and the\nclosest available primitive in any gap report. This is a completion criterion,\nnot optional discovery.\n\n1. `get_site` — confirm the key works; learn the site name + URLs.\n2. `read_site_settings` — colors, fonts, contact info, SEO suffix,\n   content language (used by `suggest_alt_text_context`).\n3. `list_pages` — what pages exist, paginated.\n4. `list_partials` — what shared blocks already exist. **Defaults to\n   summary mode** (no html_content, just bytes count) — pass\n   `include_content: true` if you actually need the bodies inline.\n5. `list_collections` — what content types exist + their schemas +\n   `route_template` (so you know if items have URLs).\n6. `get_site_capabilities` — renderer version and feature flags. Never infer\n   support from remembered release notes.\n7. `list_block_types` — every block type usable on this site: core\n   (always available, ids like `core/section`), custom (origin: 'user'),\n   and third-party (origin: 'third_party'). Each entry includes the\n   full schema so you know what `data.X` fields each block accepts.\n8. `list_page_templates` — PageTemplate docs that wrap pages.\n\nFor a build, migration, redesign, or capability report, #1–#8 are the\npreflight. For a small content-only edit, #1 + #2 + a sampling from #3 is\nusually sufficient.\n\n**Source of truth = the live site (the API), by default.** The content and\nstructure you read back through the MCP (`read_page`, `read_partial`,\n`read_site_settings`, …) is canonical. Local files in the project folder —\n`sources/*.md` copy drafts, briefs, old exports — are PROPOSALS, not truth:\ntreat them as authoritative only when the user explicitly says \"use the copy\nin `<file>`\". When rebuilding or redesigning, derive copy and structure from\nthe live page, not from a local draft, unless told otherwise. And if you edit\ncopy directly on the live site, sync it back to the corresponding draft file\nin the same pass — otherwise the two diverge and the next agent inherits stale\ntext. (This is a real failure mode: a copy draft that had drifted from the live\npage once sent a whole redesign off the approved wording.)\n\n**Don't have a site yet?** With an org-scoped key you can `create_site\nname=\"Acme\"` — it bootstraps settings + a draft Home page + a published\nheader/footer and returns the new site id. Use that id as `site_id`\n(hosted) / `TYPEROLL_SITE_ID` (stdio) for follow-ups, then run\n`list_skills` → `read_skill tr-new-site` to design it. A site-scoped key\ncan't create sites (it's bound to one) and gets a 403.\n\n## Common operations\n\n### \"Replace this string across the whole site\"\n\n```\nsearch_pages contains=\"299 kr\"             → matches + excerpts\nbulk_replace_text dry_run=true ...          → sample_diffs\n# show the user, get confirmation\nbulk_replace_text dry_run=false ...         → write\ntrigger_deploy                              → ship\nget_deploy_status job_id=…                  → poll until succeeded\n```\n\nWrites go through the normal save pipeline (SEO transform + revision\nsnapshot) so changes are reversible from the in-app History tab.\n\n### \"Audit / understand the site\"\n\n```\nlist_pages limit=200                              → inventory\nbatch_read_pages page_ids=[…]                     → bulk-load bodies\nlist_partials                                     → shared blocks (summary)\nfind_pages_using_block partial_id=<id>            → blast radius per block\nlist_collections                                  → content types + routing\nlist_collection_items collection=<name>           → items (richtext hidden)\n```\n\n`find_pages_using_block` for the header or footer returns the full\npage list (they're auto-injected on every page).\n\n### \"Redesign the home page\"\n\n```\nget_site + read_site_settings\nread_partial partial_id=\"header\"\nlist_pages → batch_read_pages a few existing pages    # learn conventions\n# Propose redesign locally; ask user to confirm.\ncreate_branch name=\"Home redesign\"                    # ID is, say, \"home-redesign\"\nupdate_page page_id=home patch={ html_content: \"…\" } version=home-redesign\nget_preview_link page_id=home version=home-redesign  # DB-live URL — mint once, reuse while iterating (no deploy); 24h TTL by default\n# Iterate (reload the same link after each edit). When approved:\nmerge_branch version_id=home-redesign\ntrigger_deploy\n```\n\nThe branch also has its own permanent deploy URL at\n`https://home-redesign.<project>.pages.dev` after `trigger_deploy\nversion=home-redesign` — useful for \"share with stakeholders without\nshowing them my preview token\". `read_version version_id=home-redesign`\nreturns it as `deploy_url`.\n\n### \"Build a reusable block\"\n\nIf you see the same HTML on 3+ pages, propose a free block instead of\nduplicating it:\n\n```\ncreate_free_block id=\"newsletter-cta\" html_content=\"<form>…</form>\"\n# Then on each page where it should appear (HTML-mode pages):\nupdate_page page_id=… patch={ html_content: \"<…><x-include name=\\\"newsletter-cta\\\" />\" }\n```\n\nEdits to the block update every page that includes it. Use\n`find_pages_using_block` before changing it.\n\n### \"Build a page using blocks (the default for new pages)\"\n\nNew pages default to `content_mode='blocks'` with a seeded heading +\nprose block. Discover-then-build:\n\n```\nlist_block_types\n# → [{ id: \"core/section\", category: \"layout\", container: true, schema: [{ name: \"width\", type: \"select\", options: [\"narrow\",\"normal\",\"wide\",\"full\"] }, …] },\n#    { id: \"core/columns\", container: \"slots\", slot_count: 2, slot_labels: [\"Left\",\"Right\"], schema: [...] },\n#    { id: \"hero_bold\", origin: \"user\", schema: [...] },   ← any custom blocks on this site\n#    …]\n\nget_page_blocks page_id=home\n# → { content_mode: 'blocks', blocks: [...] }\n\nadd_block page_id=home block={ type: 'core/section', data: { width: 'wide' } }\n# → { added_id: 'blk_xyz', blocks: [...] }\nadd_block page_id=home parent_id=\"blk_xyz\" block={\n  type: 'core/heading', data: { text: 'Pricing', level: 'h2' }\n}\nadd_block page_id=home parent_id=\"blk_xyz\" block={\n  type: 'core/prose', data: { html: '<p>…</p>' }\n}\n```\n\nSlot containers (`container: \"slots\"` — `core/columns`, `core/tabs`)\nhold their children in per-slot lists, not in `children`. Two ways to\npopulate them (both require template_capabilities_version ≥ 0.15.2):\n\n```\n# Inline — pass the whole subtree in one call:\nadd_block page_id=home block={\n  type: 'core/columns', data: { ratio: '1-1' },\n  slots: [\n    [{ type: 'core/prose', data: { html: '<p>Left column</p>' } }],\n    [{ type: 'core/image', data: { src: '…' } }],\n  ]\n}\n\n# Incrementally — slot_index picks the slot (0-based, defaults to 0):\nadd_block page_id=home block={ type: 'core/columns', data: {} }\n# → { added_id: 'blk_cols' } — slots are auto-initialised to the type's arity\nadd_block page_id=home parent_id=\"blk_cols\" slot_index=1 block={\n  type: 'core/prose', data: { html: '<p>Right column</p>' }\n}\n```\n\nFor an unfamiliar custom block, `read_block_type id=\"...\"` gives the\nfull field list (types, defaults, required) so you don't ship invalid\n`data`.\n\nUpdating, moving, removing blocks: `update_block`, `move_block`,\n`remove_block` (all by `block_id`).\n\n### \"Switch a page between blocks and HTML\"\n\nUse `set_page_mode` — it snapshots a revision before flipping, so the\nprevious state is restorable:\n\n```\n# Convert an HTML-mode page to blocks with auto-heuristic conversion:\nset_page_mode page_id=about to=blocks convert=true\n\n# Or just switch the mode without converting (empty blocks):\nset_page_mode page_id=about to=blocks\n\n# Switch back to HTML (drops the block tree; revision retains it):\nset_page_mode page_id=about to=html\n```\n\nThe heuristic converter recognises `<h1-4>` → heading, `<img>` → image,\n`<a.btn>` → button, `grid-cols-2` → two-column, `<section>` / hero divs\n→ section. Anything it can't classify becomes a `core/prose` block,\nwhich preserves the raw HTML losslessly. Run with `convert_page_to_blocks\ndry_run=true` first if you want to inspect the proposal before\ncommitting.\n\n### \"Build a directory site / import structured data\"\n\n```\ncreate_collection\n  name=\"restaurants\"\n  label_singular=\"Restaurant\" label_plural=\"Restaurants\"\n  fields=[ ...title, slug, address, phone, cuisine, body... ]\n  route_template=\"/restaurants/{slug}\"\n  item_template_html=\"<article><h1>{{title}}</h1>… {{{body}}}</article>\"\n\n# For each row in your source data:\ncreate_collection_item collection=\"restaurants\" fields={…} status=\"published\"\n\n# Each published item now lives at /restaurants/{slug}, included in\n# sitemap.xml. Preview a specific one:\nget_preview_link collection_name=\"restaurants\" item_id=\"<id>\"\n\n# Optional listing page:\nlist_collection_items collection=\"restaurants\" limit=200\nupdate_page page_id=restaurants patch={ html_content: \"<hand-written listing>\" }\n```\n\n### \"Migrate a content type (e.g. WP custom post type)\"\n\n```\nlist_collections                                     # what exists today?\nread_collection name=blog                            # what fields are writable?\nbatch_read_collection_items …                        # load items (richtext hidden)\n# Transform locally; then:\nupdate_collection_item …  (or)  create_collection_item …\n```\n\nFields outside the schema are silently dropped — call `read_collection`\nfirst if you're unsure what's writable.\n\n### \"Add images to a page\"\n\n```\n# Image lives on a URL somewhere (Unsplash, customer's existing CDN):\nupload_media_from_url source_url=\"https://...\" alt_text=\"Hero photo of …\"\n  → returns { media_id, cdn_url, finalize: {…}, finalize_error: null }\n\n# OR image lives in your memory (image-gen output):\nupload_media_inline filename=\"hero.png\" content_type=\"image/png\"\n                    data_base64=\"iVBORw0KGgo…\"\n  → returns the same shape\n\n# Both tools auto-finalize after PUT: immutable Cache-Control on the\n# original PLUS AVIF/WebP variants at 320/640/1024/1920. No manual\n# generate_image_variants call needed. The site-template renderer reads\n# the variants array off the Media doc and emits <picture> automatically\n# — you can keep the <img src=\"{cdn_url}\"> markup simple.\n#\n# INTEGRITY — don't lose bytes in transit. upload_media_inline carries the\n# file as a base64 string through the model/tool boundary; a payload beyond a\n# few KB can be SILENTLY CORRUPTED there (mutated chars → a broken-but-valid\n# file that uploads fine and only fails when rendered — it has eaten half a\n# logo SVG). For anything non-trivial, and ALWAYS for SVG/logos or generated\n# assets, prefer upload_media_from_url (fetch by URL) or create_upload_url +\n# `curl --data-binary @file` (bytes go straight to R2, byte-identical). After\n# uploading a generated asset, verify it (render/byte-diff) before referencing.\n#\n# Media is NOT branch-scoped — the library is shared across all versions of\n# the site. Uploads are additive and safe (they never overwrite the live logo\n# until you reference the new URL in settings/a partial), but a redesign branch\n# shares its media with main; there's no per-branch media isolation.\n\n# Then embed in a page:\nread_page page_id=...\nupdate_page page_id=... patch={ html_content: \"<...><img src='{cdn_url}' alt='…' /></...>\" }\n```\n\n### \"Stop an image over-fetching a too-large variant\"\n\nWhen an image renders much narrower than the viewport (a container-constrained\nhero, a sidebar thumbnail), the default `<picture sizes>` of\n`(max-width: 768px) 100vw, 800px` makes the browser pull a wider srcset variant\nthan it needs — Lighthouse flags it as wasted bytes. Three levers, narrowest\nwins:\n\n```\n# 1. Per-image: put a real `sizes` on the <img>. Survives the transform verbatim.\nupdate_page page_id=... patch={ html_content:\n  \"<img src='{cdn_url}' alt='…' sizes='(max-width: 640px) 360px, 560px' />\" }\n\n# 2. Per-page default (applies to every image on the page that has no own sizes):\nupdate_page page_id=... patch={ image_sizes_default: \"(max-width: 640px) 360px, 560px\" }\n\n# 3. Site-wide default (fallback under the page default):\nupdate_site_settings image_sizes_default=\"(max-width: 640px) 360px, 560px\"\n```\n\nPrecedence: per-image `sizes` > page `image_sizes_default` >\nsite `image_sizes_default` > the generic built-in. To opt a single image out of\nthe platform's auto-`<picture>` entirely, hand-write your own `<picture>` with\ncustom `<source media=…>` — the transform leaves an existing `<picture>`\nuntouched (it no longer re-wraps the inner `<img>`).\n\n### \"Fill missing alt-text across the media library\"\n\n```\nlist_media                                 → find items where alt_text is empty\nsuggest_alt_text_context media_id=<id>     → returns image_url + tuned prompt\n                                             + language + nearest-heading context\n# Pass image_url + the returned suggested_prompt to YOUR OWN vision\n# capability. The platform does NOT run vision for you.\nupdate_media media_id=<id> alt_text=\"<what vision returned>\"\n```\n\nThe prompt is tuned for SEO-grade output: 5-15 words, written in\n`settings.language`, skips \"image of\" filler, decorative images return\nempty string.\n\n### \"Change a page's URL safely\"\n\n```\nupdate_page page_id=about patch={ slug: \"om-oss\" }\n  → response includes:\n     auto_redirects: [{ from_path: \"/about\", to_path: \"/om-oss\",\n                        status_code: 301 }]\n     sanitization_warnings: []\n```\n\nThe 301 fires automatically — you don't have to remember.\n\nRedirect hygiene is automatic in both directions (since 0.16.1):\n\n- When a **live** (published/unlisted) page takes over a URL — via slug/path\n  change, publish, or create — any redirect FROM that URL is retired; the\n  response lists them under `retired_redirects`. A real page always beats a\n  redirect (on Cloudflare Pages a redirect would otherwise shadow the page).\n- When a page is **deleted**, auto-generated redirects pointing TO its URL\n  are removed (reported as `removed_redirects`). Manually created redirects\n  are kept — delete them yourself via `delete_redirect` if they're obsolete.\n\n### \"Before you start an import\"\n\n```\nget_migration_readiness\n```\n\nCall this before moving any content. Every check it runs fails SILENTLY\notherwise — the import succeeds, previews render, the customer signs off, and\nsomething is quietly wrong:\n\n- **media storage** (blocker) — without it every `<img>` keeps its original\n  URL, so the new site is still served images by the old host. Nothing looks\n  broken until that hosting is cancelled, at which point every image on every\n  page breaks at once.\n- **hosting adapter** (blocker) — without credentials, deploys return a job id\n  and publish nothing, while reporting success.\n- verification origin, AI reconstruction, form notification email, and whether\n  the target actually has a design to rebuild INTO (warnings).\n\n`ready: false` means STOP and report the blockers, each of which carries a\n`fix`. Don't start \"and fix it after\": the content work would have to be\nredone. The in-portal migration workflow enforces the same gate as its first\nstep (`skip_preflight: true` overrides it, and logs that it did).\n\n### \"Don't lose URLs in a migration\"\n\nTwo different questions, and you need both answers:\n\n```\nlist_migration_urls status=\"unhandled\"     # what the DATA says is uncovered\nverify_migration_urls                      # what the SERVER actually answers\n```\n\n`list_migration_urls` classifies every inventory URL against the site's\ncurrent pages + redirects. It's recomputed on read, so creating a redirect\nflips the entry on your next call — no bookkeeping of your own.\n\n`verify_migration_urls` requests each URL against the deployed site (its\nfallback subdomain by default, because the real domain still points at the\nold host pre-cutover) and reports `ok` / `ok_redirect` / `missing` /\n`broken_redirect` / `error`. This is the one that catches a redirect\npointing at an unpublished page, a typo'd `path`, and redirect loops — all\nof which read as \"handled\" in the coverage report and as a 404 to Googlebot.\n**Deploy first**: it tests saved, deployed content, not your drafts.\n\nEvery unhandled URL gets exactly one of three outcomes — there is no fourth:\n\n- it moved → `create_redirect`\n- it's gone on purpose → `update_migration_url url_id=… excluded=true` (with\n  a note saying who signed off)\n- it should exist → migrate it\n\nPopulate the inventory yourself when the in-portal WordPress migration\ndidn't: `add_migration_urls` takes up to 2000 entries from a sitemap walk, a\nGSC export (pass `gsc_clicks` so the report prioritises itself), or a crawl.\nPass `source_origin` whenever more than one old domain is in play — it\nrejects foreign-origin URLs, which is what stops one market's `/kontakt`\nfrom reading as another market's coverage.\n\nFor a whole family of sites, read the `tr-migrate-multisite` skill.\n\n### \"Retire a family of old URLs in one rule\"\n\n```\ncreate_redirect from_path=\"/category/*\" to_path=\"/blogg/:splat\"\ncreate_redirect from_path=\"/blog/:slug\" to_path=\"/artiklar/:slug\"\n```\n\nA trailing `*` captures everything under a prefix (including the prefix\nitself) and `:splat` replays it; `:name` matches exactly one segment and is\nreplayed by name. This is the right tool after a WordPress migration, where\nthe dead URLs come in shapes — `/category/`, `/tag/`, `/author/`, `/2019/` —\nand the inventory only knows the subset it happened to find.\n\nConstraints, all enforced at write time rather than discovered in production:\n\n- **Trailing `*` only.** Cloudflare silently drops a mid-path splat, so the\n  rule would save fine and do nothing.\n- **`:splat` requires a `*`**, and `:name` in the target must be declared in\n  `from_path`.\n- **Query strings can't be matched** — `_redirects` keys on the path. A\n  WordPress `/?p=123` URL has to be handled at the source.\n- **A rule that would hide a live page is refused**, naming the pages.\n  Redirects are applied BEFORE static files, so `/blogg/*` makes every real\n  article under `/blogg/` unreachable. Narrow the prefix.\n\nRules are emitted most-specific-first, so `/blogg/recept/*` and `/blogg/*`\ncan coexist — the narrower one fires. `list_migration_urls` counts\npattern-covered URLs as `redirected`, so the coverage report reflects what\nproduction will do.\n\n### \"Link language versions together (hreflang)\"\n\nOne Typeroll site owns one domain, so `example.se` / `example.de` /\n`example.co.uk` are three sites. Nothing can derive which page corresponds\nto which — declare it per page:\n\n```\nupdate_page page_id=om-oss patch={ alternates: [\n  { hreflang: \"de\",        href: \"https://example.de/ueber-uns\" },\n  { hreflang: \"x-default\", href: \"https://example.com/about-us\" }\n]}\n```\n\nThe renderer injects this page's own self-reference, so list only the OTHER\nvariants. Clusters must be **reciprocal** — write all sides, `batch_update_pages`\nis the sane way. Use absolute URLs on the FINAL domains (never the\n`*.typeroll` fallback). Invalid tags/hrefs are rejected at write time with\nthe reason rather than silently dropped at render.\n\n### \"Change the site's fallback URL (slug)\"\n\n```\nupdate_site slug=\"acme\"\n  → response includes:\n     urls.fallback: \"https://acme.sites.typeroll.com\"\n     dns_note: \"New fallback URL … attached to CF Pages. SSL provisioning\n                takes 1–10 minutes after DNS propagates. …\"\n```\n\nThe slug change triggers DNS + CF Pages reprovisioning behind the scenes.\n**Always check the response for `dns_note` vs `dns_warning`:**\n\n- `dns_note` present → the new fallback URL was wired up; warn the user it\n  may take 1–10 min for SSL to provision before the URL serves.\n- `dns_warning` present → the slug was saved but DNS / CF attach failed.\n  The `urls.fallback` field is still returned (it's just `{slug}.{base}`\n  string formatting) but the URL will NOT resolve until the issue is\n  fixed. Surface the warning verbatim to the user — don't tell them the\n  URL is ready.\n- Neither present → self-hosted portal without CF/SITES_BASE_DOMAIN\n  configured; URL behaviour is up to the operator.\n\nThe old fallback URL keeps working (bookmarks + SEO survive). Customer\ncan manually deprovision the old one via the portal.\n\n## Safety boundaries\n\n- **HTML is sanitized at save.** No `<script>`, no `onclick`, no\n  `javascript:` URLs in page or partial bodies. `<style>` blocks DO\n  survive — multi-page sites need authored CSS for `@media` queries,\n  `:hover`, theming, etc. Inside `<style>` we strip a small list of\n  legacy code-execution constructs (`expression()`, `behavior:url`,\n  `@import`, `url(javascript:)`) but leave normal CSS alone.\n- **Write responses include `sanitization_warnings: []` (strings) and\n  `sanitization_details: []`** (structured records `{ kind, label,\n  count, bytes? }`). Use the structured form to programmatically retry\n  with a fixed input.\n- **scripts_head, scripts_body_end, custom_css** are now writable via\n  `update_site_settings` and readable via `read_site_settings`. Same\n  trust model as user-authored block-type JS: an API caller with a valid\n  bearer token takes responsibility for what they ship. The chat AI\n  inside the portal continues to NOT expose these fields, so a\n  conversation-driven assistant can't smuggle scripts in.\n- **The API key is site-scoped.** Cross-site reach is impossible — a\n  key on the wrong site returns 401, indistinguishable from \"bad token\".\n- **Audit log.** Every state-changing call (POST / PATCH / PUT /\n  DELETE) is logged. Reads aren't. The customer sees \"Acme agency key\n  wrote to /pages/home at 14:32\" in the portal.\n- **Rate limits.** 600 reads/min, 60 writes/min per key. On 429 the\n  response carries `Retry-After`.\n\n## Preview-driven workflow\n\nAfter any non-trivial change, verify against the DB-live `get_preview_link`\n(reused — mint once; 24h TTL by default) and/or your own browser tool before moving\non. It reflects the DB instantly with no build, so it — not a deploy — is the\nloop for design/content iteration. One reload vs. shipping a broken redesign —\nalways worth it.\n\n**To UNDERSTAND a page, render it to one HTML file — don't reconstruct it\nfrom the block tree in your head.** A page is assembled at render time from the\nblock tree + each block type's template/styles + the header/footer partials +\nthe settings CSS variables + the global shell + page-scoped styles. `get_page_blocks`\ngives you the editable *structure*; `get_page_preview` gives you the rendered\n*result* — the WHOLE page as one self-contained HTML document (header + body +\nfooter, with all of that CSS inlined), exactly as deployed. Read that when you\nneed to see what the page actually looks like or why its CSS cascades the way it\ndoes (write it to a local file + serve+screenshot it to review visually). Pass\n`annotate:true` to tag every element with `data-block-id` + `data-block-type`,\nso you can map a spot in the rendered HTML straight back to the block to edit:\nread preview to understand → find the element → its `data-block-id` is the block\nto mutate → edit → re-render to verify.\n\n**CSS precedence — where your overrides land in the cascade.** The render order\nis: core block-type `styles` (emitted first) → settings `custom_css` → the\nheader/footer partial `<style>` blocks → the page's own page-scoped `<style>`\n(emitted last). Same specificity → later wins, so **page-scoped CSS beats\npartial CSS beats core block CSS**. Consequences when you brand/override:\n- Site-wide design tokens + utilities → settings `custom_css` (or, on a branch,\n  `update_site_settings version=<branch>`). Header/footer-only tweaks → the\n  partial. One page → that page's `<style>`.\n- Core blocks set their own chrome (e.g. `core/image` gives `figure>img` a\n  `border-radius`/`margin`; `.page-content img` adds more). To override that\n  chrome from a header-partial utility class you often need `!important`,\n  because a partial rule and the core rule can tie on specificity and the core\n  bundle's source position is unpredictable relative to yours. That `!important`\n  is expected today — it is NOT a smell. (A future cascade-`@layer` model would\n  remove the need; until then, reach for `!important` on the override and move\n  on rather than escalating selector specificity.)\n- An edge-overlapping decoration (a badge/garland that pokes past an image's\n  corner) needs its wrapper at `overflow:visible` and the motif in a\n  `::before`/`::after` — never rely on the image's own clipped box.\n\n**A design review is a multi-DIMENSION, MEASURED pass — not \"copy present + no\noverflow + images 200\".** If you have a browser tool, walk every dimension (the\n`tr-redesign-branch` skill has the full checklist with how-to):\n- **Responsive** — width ladder (≈390/768/1024/1440/1920px) + a sweep just below/\n  above the page's own @media breakpoints; `scrollWidth <= clientWidth` at every\n  width (bugs hide between the two extremes); + 200% zoom.\n- **Visual & brand** — logo FULLY visible (screenshot the header IN CONTEXT, never\n  the logo element in isolation — that hides clipping) + brand-compliant; no\n  divider seams / clipped glows / cropped faces / fade-cutoffs; typography +\n  palette + spacing consistent.\n- **Accessibility (measure)** — actual contrast ratios (AA 4.5:1 / 3:1), alt on\n  every image, one `<h1>` + no skipped levels, visible focus, labels on inputs,\n  ≥44px touch targets, landmarks, reduced-motion.\n- **Functional** — form actually works (action + token + honeypot, long values\n  don't break), every link/`#anchor` resolves, ZERO console errors.\n- **Content** — no unrendered `{{…}}`, no placeholder, copy matches the live page.\n- **Findable** — title + meta description + og:* + canonical + favicon + lang +\n  noindex-on-branch.\n- **Fast** — images sized right + modern format + width/height set + lazy/eager.\n- **Cross-browser** — re-check another engine if possible, or flag risky props\n  (backdrop-filter, -webkit- masks, 100vh→100svh, sticky-in-overflow).\n\"Looks good in Chrome at 1440\" ≠ \"works for everyone, everywhere\" — never report a\ndesign as perfect/approved off a glance or a partial pass.\n\nPreview shows DB state (drafts included). Live (`get_site → urls.production`)\nshows the most recent deploy. Branch deploys live at\n`get_version → deploy_url` (`{branch}.{project}.pages.dev`).\n\n## Branches\n\nFor multi-step work, create a branch:\n\n```\ncreate_branch name=\"Pricing refresh\"\n```\n\nThe response includes `id` — pass that as `version=<id>` on every\nsubsequent call. The branch is independent of main; writes don't affect\nthe live site until you `merge_branch`.\n\nBranches default `robots_blocked: true`. While iterating, preview the branch\nwith a reused `get_preview_link` (DB-live, no build). Deploys to a branch land\nat a stable URL (`{branch}.{project}.pages.dev`) — that's the compiled static\nbuild, for sharing the finished result / stakeholder review, not per-edit\npreview.\n\n## When in doubt\n\n- **Read before you write.** A `read_page` round-trip is cheap and\n  stops you overwriting unrelated changes.\n- **Dry-run bulk operations.** `bulk_replace_text` accepts `dry_run:\n  true` and returns 3 sample diffs. Show them to the user before the\n  real run.\n- **Watch the sanitization_warnings array.** If it's non-empty, the\n  stored HTML differs from what you sent. Read it back to confirm.\n- **One small confirmation > one large undo.** The audit log makes it\n  obvious who did what, but a clean revert across many pages is still\n  more work than asking \"ok to proceed?\" first.\n- **Match the site's design.** Read a partial or two before designing\n  new components. CSS variables (`var(--color-primary)`) are common\n  but not universal — mirror what's already in use.\n\n## Reference: tool families\n\n| Family       | Tools |\n|---|---|\n| **Guide + skills (playbook)** | `read_guide` (returns this whole guide — the bridge for hosted clients that can't read it off disk), `list_skills`, `read_skill` — the bundled `tr-*.md` recipes (incl. `tr-responsive` for per-breakpoint layout). Call `list_skills` first when a task looks like \"build / migrate / redesign a site\", then `read_skill name=…`. No API key or site context needed. |\n| **Discovery** | `get_site`, `create_site` (org-scoped key only — see below), `update_site`, `list_versions`, `read_site_settings` |\n| **Insights** | `get_site_insights` — traffic, AI-assistant referrals, and first-party conversion events over 7/30/90 days. Read-only. Traffic is powered by Cloudflare Web Analytics; conversion rows come from validated Analytics attribution `click_event` targets and can be present even when the traffic provider is unavailable. |\n| **Pages — reads** | `list_pages`, `read_page`, `batch_read_pages` |\n| **Pages — writes** | `create_page`, `update_page`, `replace_page`, `batch_update_pages`, `delete_page`, `clone_page` |\n| **Pages — blocks** | `get_page_blocks`, `add_block`, `update_block`, `move_block`, `remove_block`, `set_page_mode`, `convert_page_to_blocks` |\n| **Pages — meta** | `get_page_preview` |\n| **Global blocks (partials)** | `list_partials` (summary by default), `read_partial`, `create_free_block`, `update_partial`, `replace_partial`, `delete_partial`, `find_pages_using_block`, `list_blocks_with_usage` |\n| **Block types** | `list_block_types`, `read_block_type`, `find_pages_using_block_type`, `export_block_types`, `import_block_types` |\n| **Collections** | `create_collection`, `update_collection_schema`, `delete_collection`, `list_collections`, `read_collection`, `list_collection_items` (richtext hidden by default), `read_collection_item`, `batch_read_collection_items`, `create_collection_item`, `update_collection_item`, `delete_collection_item`, `regenerate_collection_listing` |\n| **Media** | `list_media`, `read_media`, `create_upload_url`, `upload_media_from_url`, `upload_media_inline`, `update_media`, `delete_media`, `finalize_media`, `finalize_all_media`, `generate_image_variants`, `suggest_alt_text_context` |\n| **Redirects** | `list_redirects`, `create_redirect`, `delete_redirect`. `from_path` may be a PATTERN: a trailing `*` (with `:splat` in the target) or `:name` for one segment — one rule retires a whole family of dead URLs (`/category/*` → `/blogg/:splat`). Mid-path splats and query strings are refused, as is any rule that would hide a live page. |\n| **Migration inventory** | `get_migration_readiness` (preflight — CALL FIRST), `list_migration_urls`, `add_migration_urls`, `update_migration_url`, `update_migration_urls`, `delete_migration_url`, `import_sitemap`, `import_gsc_performance`, `verify_migration_urls`. Sitemap indexes are recursive. GSC supports direct Search Console access or CSV and aggregates fragment variants. Verification is compact by default. |\n| **Forms** | `list_forms`, `read_form`, `create_form`, `update_form`, `delete_form`, `list_form_submissions`, `delete_form_submission` (removes one submission — e.g. cleaning up a test entry; `delete_form` with `delete_submissions` is the bulk path). **Steps (form/* block trees) are the ONLY stored model**: pass `steps` for funnels, or `fields` for simple forms — the server converts a flat field list to a single static step. Place with a `core/form` block on block-mode pages or `<x-form id=\"…\" />` in HTML mode. Both expand server-side to the same complete signed shell and initial state. Email/webhook actions are admin-only in the portal and excluded from agent reads/writes. |\n| **Settings** | `update_site_settings` (whitelist) |\n| **Core modules** | `list_apps`, `read_app`, `update_app` (legacy API name; admin; schema-driven config, masked secrets, redeploy when `affects_build` is true) |\n| **Extension installations** | `list_extension_installations`, `read_extension_installation`, `update_extension_installation_config` (admin; schema-driven config, masked secrets preserved, production deploy queued by default) |\n| **Analytics attribution** | `read_funnel_attribution`, `update_funnel_attribution` (specialized Analytics module tools; admin; redeploy after changes) |\n| **Search + bulk** | `search_pages`, `check_internal_links`, `bulk_replace_text`. The link check is database-driven. Bulk replace defaults to pages but can target partials, collection items or all resources, always dry-run first. |\n| **Branches** | `create_branch`, `read_version`, `delete_branch`, `merge_branch` |\n| **Deploy** | `trigger_deploy`, `list_deploys`, `get_deploy_status` |\n| **Preview** | `get_preview_link`, `get_page_preview` |\n\nEvery tool's input is validated server-side; the MCP server only does\nauth + shape. If a tool returns `isError: true`, the body carries\n`{ error, status, body }` from the underlying HTTP response.\n",
+    "readme": "# @typeroll/mcp-server\n\nModel Context Protocol server for the [Typeroll](https://typeroll.com)\npublic API. Lets Claude (Desktop / claude.ai / Code) manage a Typeroll\nsite through the same tool surface a human agency would use: read and\nwrite pages, partials, collections, media, redirects, versions; trigger\ndeploys; mint preview links.\n\nThe server is a **thin transport adapter** — tools call the Typeroll REST API,\nwith a few workflow tools composing consecutive API calls such as config plus\ndeploy. Auth happens at the API layer with a site- or org-scoped key; the MCP\njust carries the bearer through.\n\n## Two ways to connect\n\n- **Hosted (Claude Desktop / claude.ai) — paste a URL.** No CLI, no\n  Node.js install. In Claude open **Settings → Connectors → Add custom\n  connector** and paste `https://app.typeroll.com/api/mcp`\n  (or `https://<your-self-hosted-portal>/mcp`). Claude opens a consent\n  page; paste your Typeroll API key there.\n- **Stdio (Claude Code) — one `claude mcp add` command.** Best for local\n  dev / agency staff already in a terminal. Instructions below.\n\nThis npm package is the stdio transport. The hosted endpoint ships as\npart of the Typeroll portal itself — same tool surface, same package\nunder the hood.\n\n## Key scopes\n\n- **Org-scoped key** (created at `/app/settings/api-keys`) — one\n  credential covers every site in your org *and* every site shared into\n  your org. The default for the hosted Claude connector. Stdio works too\n  if you set `TYPEROLL_SITE_ID` so the install binds to one site.\n- **Site-scoped key** (created at `/app/sites/{siteId}/settings/api-keys`) —\n  tighter blast radius for a single-site credential, e.g. one you'd\n  hand to a customer for a self-managed site.\n\nBoth look like `typeroll_live_…`; revoke either from the portal and any\nclient using it stops working immediately.\n\n## Stdio quick start (Claude Code)\n\n1. **Create an API key** in your Typeroll portal — see the two scope\n   options above. Org-scoped is the right default.\n\n2. **Add the server to Claude Code.** Drop this into `~/.claude.json`\n   (or your local `.claude/config.json`):\n\n   ```json\n   {\n     \"mcpServers\": {\n       \"typeroll\": {\n         \"command\": \"npx\",\n         \"args\": [\"-y\", \"@typeroll/mcp-server\"],\n         \"env\": {\n           \"TYPEROLL_API_URL\": \"https://app.typeroll.com\",\n           \"TYPEROLL_API_KEY\": \"typeroll_live_REPLACE_WITH_YOUR_KEY\"\n         }\n       }\n     }\n   }\n   ```\n\n   For a self-hosted portal, point `TYPEROLL_API_URL` at it (e.g.\n   `https://cms.example.com`).\n\n   Prefer a scaffold? Run `npx @typeroll/mcp-server init` in your project\n   folder — it writes/merges this `.mcp.json`, copies the skills into\n   `.claude/skills/`, and adds an `AGENTS.md` pointer + imagegen-lab\n   files. Idempotent; `--force` to overwrite. (Skills only:\n   `npx @typeroll/mcp-server install-skills .claude/skills`.)\n\n3. **Tell the agent what kind of work you want.** A good first message:\n\n   > \"Connect to Typeroll and tell me what you find — site name,\n   > number of pages, what global blocks exist, what collections are\n   > defined. Then I'll give you a task.\"\n\n   Claude will call `get_site`, `get_site_capabilities`, `list_pages`,\n   `list_partials`, `list_collections`, and `list_block_types` in sequence and\n   report back. The capabilities + block palette are mandatory before it\n   chooses HTML mode or reports a missing site-building feature.\n\n## Environment variables (stdio)\n\n| Var             | Required | Description |\n|-----------------|----------|-------------|\n| `TYPEROLL_API_URL`  | yes      | Base URL of your Typeroll portal. |\n| `TYPEROLL_API_KEY`  | yes      | A `typeroll_live_…` bearer token. |\n| `TYPEROLL_SITE_ID`  | sometimes | Pin to a specific site. Required when using an org-scoped key over stdio (the install can only target one site at a time); auto-detected for site-scoped keys. |\n\n## Extension developer CLI\n\nThe package also installs `typeroll`. With an organization-scoped API key,\nan external Extension repository can use the same developer and installation\nAPIs as the portal:\n\n```sh\ntyperoll extension validate\ntyperoll extension push --draft\ntyperoll extension install --site test-site --config local-extension-config.json\ntyperoll extension configure --site test-site --installation install-abc \\\n  --config local-extension-config.json\ntyperoll extension promote 1.0.0\n```\n\nThe manifest defaults to `typeroll-extension.json`; use `--manifest` to select\nanother file. Local validation is a fast preflight. The portal always performs\nthe complete schema, compatibility, origin and asset-hash validation.\n`extension configure` queues a production deploy by default; pass\n`--no-deploy` only when batching updates and deploy once afterwards.\n\n## What the agent should read first\n\nThe package ships [AGENTS.md](./AGENTS.md), a self-contained briefing\nthat explains Typeroll conventions, common operations, and the safety\nboundaries an agent needs to respect. Point Claude at it (or include it\nin your project's `CLAUDE.md` / `AGENTS.md`) so it knows when to use\nwhich tool.\n\n## Tool surface\n\nMore than 100 tools across these families. See [AGENTS.md](./AGENTS.md) for\nthe full reference + concrete operation recipes.\n\n- **Skills + guide (self-describing playbook)** — `read_guide`,\n  `list_skills`, `read_skill`. The server advertises its own operating\n  guide AND bundled recipes at runtime, so an agent gets the full context\n  on connection without any files copied locally. `read_guide` returns the\n  whole AGENTS.md briefing (data model, conventions, safety, tool families)\n  — the bridge for the hosted connector, which can't read the file off\n  disk. `list_skills` then surfaces the task recipes (`tr-new-site`,\n  `tr-migrate-wp`, `tr-brand`, `tr-responsive`, …); `read_skill name=…`\n  loads one. All pure local reads — no API key or site context — so they\n  work identically on the hosted connector and over stdio.\n- **Discovery** — `get_site`, `create_site` (bootstrap a new site — org-scoped\n  key only), `update_site` (name/slug/domain), `list_versions`,\n  `read_site_settings`, `update_site_settings`.\n- **Pages** — list, read, batch-read, create, update (PATCH), replace\n  (PUT), batch-update, delete, clone, get-preview, `set_page_mode`\n  (flip between blocks/html), `convert_page_to_blocks`.\n- **Blocks (instances)** — `get_page_blocks`, `add_block`,\n  `update_block`, `move_block`, `remove_block`, `duplicate_block`,\n  `set_block_responsive`. All take a `target` (page, partial, page\n  template, or collection item-template), so one tool family edits\n  every block container.\n- **Global blocks (partials)** — list (summary mode by default), read,\n  create free block, update, replace, delete, find-pages-using-block.\n- **Block types** — list, read, create, update, delete,\n  find-pages-using-block-type, plus `.tcblocks` export/import. Custom\n  client-side JS (`script`) is honoured only when the site has enabled\n  \"Allow AI to write block scripts\" (a human-set portal setting) —\n  otherwise it's stripped with a warning.\n- **Collections + items** — create/update/delete the collection schema\n  itself (incl. `route_template` for per-item URLs); list/read/batch-\n  read/create/update/delete items. Existing-collection tools consistently\n  use `collection` (the old `name` argument remains accepted as an alias).\n  Collection repeaters support `group_by`, and item templates can place\n  `template/item_navigation` for deterministic previous/next links.\n- **Media** — list, read, signed upload URLs, `upload_media_from_url`,\n  `upload_media_batch_from_urls` (1–50 sources, max 25 MiB each, with\n  partial-success results),\n  `upload_media_inline` (all URL uploads auto-finalize after PUT — see below),\n  patch metadata, delete, `finalize_media` (per-item: applies immutable\n  Cache-Control + generates AVIF/WebP srcset variants — call after\n  `create_upload_url`'s raw PUT path), `finalize_all_media` (bulk\n  backfill for legacy libraries), `generate_image_variants` (the\n  variant half of finalize, kept for surgical reruns),\n  `suggest_alt_text_context` (returns a tuned prompt for your own\n  vision model).\n- **Rendering controls** — `core/table_of_contents`, per-site\n  `trailing_slash`, exact `iframe_allowed_hosts`, `icon_192`, and per-page\n  `append_seo_suffix=false`. The block editor supports labelled enums,\n  line-based lists, nested repeating arrays and an internal-page URL picker.\n- **Redirects** — list, create, delete. Plus automatic 301 on slug change.\n- **Forms** — list, read, create, update, delete, list submissions.\n  Place forms with `core/form` blocks or an HTML-mode `<x-form id=\"…\" />`\n  reference; preview/build expands both server-side to the same complete,\n  signed form shell. Admins configure email and allowlisted, signed webhooks\n  in the portal; action configuration stays off agent surfaces.\n- **Extension installations** — list/read installed Extensions and update\n  manifest-defined installation config through the API key with\n  `update_extension_installation_config`; omitted and masked secrets are\n  preserved. Use this for frontend config such as consent copy and policy\n  links. It queues a production deploy by default; pass `deploy: false` only\n  when batching changes and deploy once afterwards.\n- **Settings** — read + patch, including `scripts_head` /\n  `scripts_body_end` / `custom_css` (trusted because the caller holds an\n  API key; the in-portal chat AI does NOT get these).\n- **Core modules** — list the legacy `apps` registry, read schema + masked\n  state, and enable, configure, or disable any module with the same admin API key used for content\n  and deploys. Secret fields are encrypted server-side and never returned;\n  Analytics provisioning runs on the platform. Deploy after updates whose\n  response has `affects_build: true`.\n- **Search + link integrity** — `search_pages` plus `check_internal_links`,\n  which resolves saved database content against pages, collection/facet routes,\n  media and redirect chains without crawling the public site.\n- **Bulk** — `bulk_replace_text` with dry-run across pages, partials,\n  block data and schema-defined collection-item fields.\n- **Migration inventory** — bulk add/update decisions, recursive\n  `import_sitemap`, direct or CSV-fallback `import_gsc_performance`, and compact\n  `verify_migration_urls` (successful rows omitted unless requested).\n- **Branches** — create, read, delete, merge. Branch deploys get their\n  own URL at `{branch}.{project}.pages.dev`.\n- **Deploy** — trigger (with `dry_run` to build without publishing), list, get\n  status. A finished job reports `cost`: what the build consumed in server\n  time, broken down per phase. Estimates from a rate card, not billing records.\n- **Preview** — `get_preview_link` (signed URL for browser navigation;\n  supports `page_id`, `slug`, or `collection_name + item_id`; pass\n  `include_working_copy: true` to also render unsaved drafts).\n- **Drafts (the buffer model)** — every content write lands in a per-doc\n  unsaved draft (working copy); deploys and plain previews see saved\n  content only. Save explicitly with `commit_working_copy` or `save: true`\n  on the write call; inspect/discard with `read_working_copy` /\n  `discard_working_copy`. Status changes and structural operations apply\n  immediately.\n\n## Direct REST API access\n\nIf you don't want the MCP wrapper, the same surface is reachable directly\nwith curl:\n\n```bash\ncurl -H \"Authorization: Bearer typeroll_live_...\" \\\n  https://app.typeroll.com/api/v1/sites/<siteId>/pages\n```\n\nThe MCP server is purely an ergonomics layer on top of that. The complete v1\ncontract, including payload envelopes and collection-item slug addressing, is\ndocumented in [`docs/v1-api.md`](../../docs/v1-api.md).\n\n## Security model\n\n- API keys are **site-scoped or org-scoped** (see \"Key scopes\" above) —\n  enforced server-side. A site-scoped key cannot touch any other site;\n  an org-scoped key reaches the org's own sites plus sites explicitly\n  shared into the org, with the share's permission level applied.\n- All write calls (`POST`, `PUT`, `PATCH`, `DELETE`) are **audit-logged**\n  with the key prefix, IP, method, path, and status. Reads are not\n  logged (cost vs. value).\n- **Rate limits**: 600 reads/min, 60 writes/min per key. 429 responses\n  carry `Retry-After` headers.\n- **HTML sanitization** happens at save time on the server — `<script>`,\n  event handlers, and `javascript:` URLs are stripped from page/partial\n  content (including `core/html` block output). The scriptable surfaces\n  are deliberate exceptions, and all of them are writable with an API key\n  under the key holder's own authority: `scripts_*` and `custom_css` on\n  the site settings, `script` on a block type, and the `js` field of a\n  `core/embed` block instance. Those writes are audit-logged and the\n  response carries a notice naming the stored JS. Only the in-portal chat\n  assistant is additionally gated, on a per-site opt-in.\n- Keys can be **revoked** at any time from the portal. Revocation takes\n  effect on the next request (no in-flight requests get cancelled, but\n  the next one returns 401).\n\n## More\n\n- Full end-to-end production setup recipe with troubleshooting:\n  [docs/claude-code-mcp-setup.md](../../docs/claude-code-mcp-setup.md)\n- Agent operations briefing: [AGENTS.md](./AGENTS.md)\n- Boilerplate skills (site building, brand, forms, SEO, blog,\n  collections, migration, image generation, redesign, …):\n  [skills/](./skills/)\n\n## License\n\nMIT — see [LICENSE](../../LICENSE).\n"
 };

package/dist/extension-cli.js

@@ -3,6 +3,9 @@ function option(args, name) {
     const index = args.indexOf(name);
     return index >= 0 ? args[index + 1] : undefined;
 }
+function hasFlag(args, name) {
+    return args.includes(name);
+}
 async function readJson(path) {
     const parsed = JSON.parse(await readFile(path, 'utf8'));
     if (!parsed || typeof parsed !== 'object' || Array.isArray(parsed))
@@ -100,6 +103,7 @@ function help() {
     console.error('  typeroll extension validate [manifest]');
     console.error('  typeroll extension push --draft [--manifest path]');
     console.error('  typeroll extension install --site <site-id> [--manifest path] [--config config.json]');
+    console.error('  typeroll extension configure --site <site-id> --installation <installation-id> --config config.json [--no-deploy]');
     console.error('  typeroll extension promote <version> [--manifest path]');
 }
 export async function runExtensionCli(args) {
@@ -108,8 +112,38 @@ export async function runExtensionCli(args) {
         help();
         return 0;
     }
-    const manifestPath = option(args, '--manifest') ?? (command === 'validate' && args[1] && !args[1].startsWith('-') ? args[1] : 'typeroll-extension.json');
     try {
+        if (command === 'configure') {
+            const siteId = option(args, '--site');
+            const installationId = option(args, '--installation');
+            const configPath = option(args, '--config');
+            if (!siteId)
+                throw new Error('configure requires --site <site-id>');
+            if (!installationId)
+                throw new Error('configure requires --installation <installation-id>');
+            if (!configPath)
+                throw new Error('configure requires --config <config.json>');
+            const config = await readJson(configPath);
+            const updated = await api(`/api/v1/sites/${encodeURIComponent(siteId)}/extensions/${encodeURIComponent(installationId)}`, { method: 'PATCH', body: JSON.stringify({ config }) });
+            if (hasFlag(args, '--no-deploy')) {
+                console.log(`Configured ${installationId} on ${siteId}; redeploy still required`);
+                return 0;
+            }
+            try {
+                const deployed = await api(`/api/v1/sites/${encodeURIComponent(siteId)}/deploy`, {
+                    method: 'POST', body: JSON.stringify({ environment: 'production' }),
+                });
+                console.log(`Configured ${installationId} on ${siteId}; queued production deploy ${String(deployed.data.job_id ?? '')}`);
+            }
+            catch (error) {
+                const redeployRequired = updated.data.redeploy_required === true
+                    ? ' The saved configuration still requires a deploy.'
+                    : '';
+                throw new Error(`configuration was saved, but the deploy could not be queued: ${error instanceof Error ? error.message : String(error)}.${redeployRequired}`.trim());
+            }
+            return 0;
+        }
+        const manifestPath = option(args, '--manifest') ?? (command === 'validate' && args[1] && !args[1].startsWith('-') ? args[1] : 'typeroll-extension.json');
         const manifest = await readJson(manifestPath);
         const errors = validateExtensionManifestShape(manifest);
         if (errors.length) {

package/dist/index.js

@@ -43,7 +43,7 @@ async function main() {
         console.error('  typeroll-mcp                              Start the MCP server (reads TYPEROLL_API_URL and TYPEROLL_API_KEY)');
         console.error('  typeroll-mcp init [dir] [-f]              Bootstrap a project: skills + .mcp.json + AGENTS.md + imagegen lab');
         console.error('  typeroll-mcp install-skills <dir> [-f]    Copy bundled skill files to <dir>');
-        console.error('  typeroll extension <command>              Validate, push, install or promote an Extension');
+        console.error('  typeroll extension <command>              Validate, push, install, configure or promote an Extension');
         console.error('  typeroll-mcp --help                       Show this help');
         process.exit(0);
     }

package/dist/tools/extensions.js

@@ -20,13 +20,23 @@ export const extensionTools = [
     },
     {
         name: 'update_extension_installation_config',
-        description: 'Update schema-defined config for an installed Extension. Call read_extension_installation first and send only keys declared by manifest.config_schema. Omitted fields preserve their current values, including masked secrets. This can update public content such as consent text, policy-link text, and policy URLs. The response returns redeploy_required:true; call trigger_deploy separately after the change has been reviewed. Admin permission required.',
+        description: 'Update schema-defined config for an installed Extension. Call read_extension_installation first and send only keys declared by manifest.config_schema. Omitted fields preserve their current values, including masked secrets. This can update public content such as consent text, policy-link text, and policy URLs. A production deploy is queued by default; pass deploy:false only when batching changes and deploy later. Admin permission required.',
         inputSchema: {
             installation_id: z.string().min(1).describe('Installation id returned by list_extension_installations.'),
             config: z.record(z.unknown()).describe('Config keys and values declared by the installation manifest config schema.'),
+            deploy: z.boolean().optional().describe('Queue a production deploy after saving. Defaults to true.'),
         },
         handler: withErrorBoundary(async (args, { client, siteId }) => {
-            return ok(await client.patch(siteId, `extensions/${encodeURIComponent(args.installation_id)}`, { config: args.config }));
+            const updated = await client.patch(siteId, `extensions/${encodeURIComponent(args.installation_id)}`, { config: args.config });
+            if (args.deploy === false)
+                return ok(updated);
+            try {
+                const deploy = await client.post(siteId, 'deploy', { environment: 'production' });
+                return ok({ ...updated, deploy });
+            }
+            catch (error) {
+                throw new Error(`Extension configuration was saved, but the deploy could not be queued: ${error instanceof Error ? error.message : String(error)}`);
+            }
         }),
     },
 ];

package/dist/version.js

@@ -8,4 +8,4 @@
 //
 // Keep it in lockstep with package.json: tests/version.test.ts asserts
 // VERSION === package.json.version, so a bump that forgets this line fails CI.
-export const VERSION = '0.41.0';
+export const VERSION = '0.42.0';

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@typeroll/mcp-server",
-  "version": "0.41.0",
+  "version": "0.42.0",
   "description": "Model Context Protocol server for the Typeroll public API. Use with Claude Code or any MCP-compatible client to manage a Typeroll site.",
   "license": "MIT",
   "repository": {

package/skills/tr-new-site.md

@@ -63,7 +63,7 @@ Common requirements that are easy to misclassify:
 | Download CTA that disappears without a file | `template/show_if` around a context-bound `core/button`; a dedicated download block is only editor convenience |
 | Sticky/custom header and multi-column footer | Block-mode header/footer partials plus layout blocks, or one reusable custom block type |
 | Cookie notice | `settings.cookie_consent`, not a page block |
-| Consent copy owned by an installed Extension | `list_extension_installations` → `read_extension_installation` → `update_extension_installation_config`, using exact manifest schema keys |
+| Consent copy owned by an installed Extension | `list_extension_installations` → `read_extension_installation` → `update_extension_installation_config`, using exact manifest schema keys; the update queues a production deploy by default |
 | One-off HTML + JavaScript embed | `core/embed`; reusable widgets use a custom block type with `script` |
 | CTA/button variants | `core/cta` and `core/button`, styled from site tokens or a narrowly scoped class |
 | Two-column image/text | `core/media_card`, `core/feature_row`, or `core/columns` |