Added value: +{
+  "description": "Plain-text files, each `{path, content}`, zipped for you: a page with its own CSS, JS, JSON or SVG files (name the entry `index.html`), or one Markdown, CSV, Mermaid, notebook, code or .tsx/.jsx React component file on its own, rendered as its kind under its path (a component needs a default export and may import only react, react-dom, recharts, lucide-react, d3, framer-motion and @radix-ui/react-*). Paths are relative to the bundle root with forward slashes (`index.html`, `assets/app.js`). Text only: an image, a font, a PDF or any binary is referenced at a URL (add_external_origins) or sent through create_bundle_upload. At most 100 entries and 2 MB of text in total (UTF-8), and 3.5 MB once JSON-escaped (a control character other than backspace, tab, newline, form feed and carriage return, `<`, `>`, `&` or any non-ASCII character counts 6 bytes, an emoji 12, and `\"`, `\\` and `/` 2, so non-ASCII text may count three times its size and markup-heavy text up to six): in practice about 2 MB of plain English or 1 MB of non-Latin or tag-dense text. The rest of the 4 MB request body is for the call itself. Like `html`, this works from any client, a chat window with no shell included.",
+  "items": {
+    "additionalProperties": false,
+    "properties": {
+      "content": {
+        "description": "The file as UTF-8 text",
+        "type": "string"
+      },
+      "path": {
+        "description": "Relative path inside the bundle, spelled exactly as the page's own references spell it (two Unicode spellings of one accented name count as one entry)",
+        "minLength": 1,
+        "type": "string"
+      }
+    },
+    "required": [
+      "path",
+      "content"
+    ],
+    "type": "object"
+  },
+  "maxItems": 100,
+  "minItems": 1,
+  "type": "array"
+}

Previous value: -"A single self-contained HTML document"New value: +"A single self-contained HTML document, as text. The first choice for one page: it needs no encoding, no shell and no network of your own. At most 3.5 MB once JSON-escaped (a control character other than backspace, tab, newline, form feed and carriage return, `<`, `>`, `&` or any non-ASCII character counts 6 bytes, an emoji 12, and `\"`, `\\` and `/` 2, so non-ASCII text may count three times its size and markup-heavy text up to six): in practice about 3 MB of plain HTML, about 2 MB when tag-dense; the rest of the 4 MB request body is for the call itself. A bigger page: Trim it, split it across more than one site, give the person the text to save as a file and hand it over, or send it through create_bundle_upload, which needs a shell that makes the SnapHost tool calls itself."

Previous value: -"Reference a bundle already uploaded via create_bundle_upload (sent with upload_bundle_part, or PUT to its `upload_url`); use this for a large or multi-file app (e.g. a built React site) that exceeds the inline `html`/`zipBase64` limit, or for a single file that is not a zip (Markdown, a notebook, a PDF, an image, a video, an audio, a CSV, a Mermaid diagram or a text file), sent as it is together with `filename`. Mutually exclusive with `html` and `zipBase64`."New value: +"Reference a bundle already uploaded via create_bundle_upload (sent with upload_bundle_part, or PUT to its `upload_url`); use this for a large or binary-carrying app (e.g. a built React site with images and fonts) that `html` or `files` cannot hold, or for a single file that is not text (a PDF, an image, a video, an audio, a .docx, .pptx or .xlsx), sent as it is together with `filename`. Sending the bytes needs a shell that makes the SnapHost tool calls itself; without a shell the person drops the file on the dashboard: the drive (https://app.snaphost.ai/) for a new site, or the site's own page (https://app.snaphost.ai/sites/<site_id>, \"Upload a new version\") to keep an existing link. Mutually exclusive with `html`, `files` and `zipBase64`."

Previous value: -"Base64-encoded .zip bundle (entry HTML + assets), for a multi-file bundle small enough to inline; a .docx, .pptx or .xlsx (zips themselves) or a one-entry zip holding a single Markdown, notebook, PDF, CSV, image, video, audio, Mermaid or text file under its real name also goes here. A single HTML document belongs in `html` instead: base64 has to be reproduced byte for byte and a long string often is not, so this input is rejected outright if the encoding or the archive checksums do not verify."New value: +"Base64-encoded .zip bundle (entry HTML + assets), for a multi-file bundle that holds binaries and fits the 4 MB request body once encoded; a .docx, .pptx or .xlsx (zips themselves) or a one-entry zip holding a single PDF, image, video or audio file under its real name also goes here. Text belongs in `html` or `files` instead: base64 has to be reproduced byte for byte and a long string often is not, so this input is rejected outright if the encoding or the archive checksums do not verify. Producing it needs a shell that makes the SnapHost tool calls itself; a client without one sends text as `files` and hands a binary to the person for the dashboard: the drive (https://app.snaphost.ai/) for a new site, or the site's own page (https://app.snaphost.ai/sites/<site_id>, \"Upload a new version\") to keep an existing link."

Previous value: -"Slug, unique within the drive the site is filed in; reuse it with the same folder_id or space to update the same site at the same URL. For a landing page the slug IS the address under /lp: pass exactly the path you were asked for; omitted, it is derived from the title and the page lands at a different address. A slug already used by a site elsewhere on the drive is refused."New value: +"Slug, unique within the drive the site is filed in; reuse it with the same folder_id or space to update the same site at the same URL. For a landing page the slug IS the address under /lp: pass exactly the path you were asked for; omitted, it is derived from the title and the page lands at a different address. A slug already used by a site elsewhere on the drive is refused, and so is one that is another site’s old address (the refusal says where that site went: update it instead of publishing a second one)."

Previous value: -"The page is built to print (a PDF that keeps the look): the viewer then offers Print, and the design check asks for print styles. Kept on later updates unless given again."New value: +"The page is built to print (a PDF that keeps the look): the design check asks for print styles, and a plain page gets the viewer's Print action (a report or a deck carries its own Download PDF button instead). Kept on later updates unless given again."

Added value: +{
+  "description": "What the site is: a report, a presentation, a dashboard, or a plain page (the default). A publish or content update of such a site runs the design check (`design_warnings`); a presentation may go full screen. Kept on later updates unless given again.",
+  "enum": [
+    "page",
+    "report",
+    "presentation",
+    "dashboard"
+  ],
+  "type": "string"
+}

Added value: +{
+  "description": "The page is built to print (a PDF that keeps the look): the viewer then offers Print, and the design check asks for print styles. Kept on later updates unless given again.",
+  "type": "boolean"
+}

Added value: +{
+  "description": "The uploaded file's own name with its extension (`report.pdf`, `photo.png`, `data.csv`). Give it once, at create_bundle_upload; a name given here as well must match it. Pass it here only for a single non-zip file whose upload was opened without one, so it is rendered as its kind; an upload that no name describes and that is neither a zip nor an HTML page is refused. For a .docx, .pptx or .xlsx sent as `zipBase64`, give its name here: it heads the bar and names the download. Ignored for a .zip bundle. A name SnapHost does not take is refused with the export that would work.",
+  "type": "string"
+}

Added value: +{
+  "description": "Whether readers may download the uploaded file itself from the page. Applies to a Word, PowerPoint or Excel file only, and is off by default for them, because the file holds what the page does not show: speaker notes, hidden slides, sheets, rows and columns, formulas, comments and tracked changes. Pass true to offer it anyway; on update_site, leave it out to keep the live version's choice. A macro-enabled .docm, .pptm or .xlsm is never offered. A CSV, image, video, audio, PDF, diagram or text file is its page and always downloads; a Markdown file or notebook becomes its page and has no download. The result reports `original_included`, and `render_notices` says what the page could not show (cells shown as formulas when a workbook carries no computed values; a file left out for size). Relay a notice to the user.",
+  "type": "boolean"
+}

Previous value: -"Reference a bundle already uploaded via create_bundle_upload (sent with upload_bundle_part, or PUT to its `upload_url`); use this for a large or multi-file app (e.g. a built React site) that exceeds the inline `html`/`zipBase64` limit. Mutually exclusive with `html` and `zipBase64`."New value: +"Reference a bundle already uploaded via create_bundle_upload (sent with upload_bundle_part, or PUT to its `upload_url`); use this for a large or multi-file app (e.g. a built React site) that exceeds the inline `html`/`zipBase64` limit, or for a single file that is not a zip (Markdown, a notebook, a PDF, an image, a video, an audio, a CSV, a Mermaid diagram or a text file), sent as it is together with `filename`. Mutually exclusive with `html` and `zipBase64`."

Previous value: -"Base64-encoded .zip bundle (entry HTML + assets), for a multi-file bundle small enough to inline. A single HTML document belongs in `html` instead: base64 has to be reproduced byte for byte and a long string often is not, so this input is rejected outright if the encoding or the archive checksums do not verify."New value: +"Base64-encoded .zip bundle (entry HTML + assets), for a multi-file bundle small enough to inline; a .docx, .pptx or .xlsx (zips themselves) or a one-entry zip holding a single Markdown, notebook, PDF, CSV, image, video, audio, Mermaid or text file under its real name also goes here. A single HTML document belongs in `html` instead: base64 has to be reproduced byte for byte and a long string often is not, so this input is rejected outright if the encoding or the archive checksums do not verify."

Previous value: -"File the new site into this folder (folder_id from list_folders). A landing page goes into the website workspace’s Marketing folder: pass that workspace’s workspace_id as well; without it the site lands in your own account and is not a landing page, whatever its slug."New value: +"File the new site into this folder (folder_id from list_folders). A landing page goes into the website workspace’s Marketing folder: pass that workspace’s workspace_id as well; without it the publish is refused on a connection that reaches several workspaces (and lands in the personal space on one that reaches only it), and the site is not a landing page, whatever its slug."

Previous value: -"Optional: the workspace (the personal space or a Pro workspace) to act in, from list_workspaces. Omit to act in the token owner's own account."New value: +"Optional: the workspace to act in, from list_workspaces (the connection reaches the person's personal space and every workspace they are a member of). Required on a write when the connection reaches more than one workspace. Omit to act in the personal space (for an older single-workspace connection, the one workspace it was authorized for)."