Orshot

List the role presets (what a clip does in a page) and look presets (visual style, each with a preview clip URL) for orshot_generate_ai_video, plus the credit rate card. Read this once before generating so you can pick a preset instead of describing style in the prompt.

orshot_get_ai_video_presets

Save a brand color. Supports hex (#FF5733, #RRGGBBAA), rgb, hsl, and CSS gradients. Use tags to categorize (e.g. ["primary", "brand"], ["background", "dark"]). Max 200 colors per workspace.

orshot_add_brand_color

Check if authenticated and list available workspaces.

orshot_auth_status

Render the same studio template multiple times with different modifications in a single call. Use this for bulk image generation — e.g. personalized certificates, product cards, social posts for a list of items. Each item in the `renders` array is a set of modifications (same format as orshot_render_template). All renders share the same template, format, and options. Smart Resize works here too: `size` (render every item at a different canvas size) and `extraSizes` (add sized copies beside each item) apply to ALL renders in the batch. Use orshot_get_studio_template_modifications first to see available parameter keys. Example: Generate 3 personalized certificates: templateId: "12345" renders: [ { "name": "Alice", "course": "Design 101" }, { "name": "Bob", "course": "Design 101" }, { "name": "Charlie", "course": "Design 201" } ]

orshot_batch_render_template

Legacy name for orshot_batch_render_template. Prefer orshot_batch_render_template — it is the primary tool and this name will be removed in a future release. Same inputs and output. Render the same studio template multiple times with different modifications in a single call. Use this for bulk image generation — e.g. personalized certificates, product cards, social posts for a list of items. Each item in the `renders` array is a set of modifications (same format as orshot_render_template). All renders share the same template, format, and options. Smart Resize works here too: `size` (render every item at a different canvas size) and `extraSizes` (add sized copies beside each item) apply to ALL renders in the batch. Use orshot_get_studio_template_modifications first to see available parameter keys. Example: Generate 3 personalized certificates: templateId: "12345" renders: [ { "name": "Alice", "course": "Design 101" }, { "name": "Bob", "course": "Design 101" }, { "name": "Charlie", "course": "Design 201" } ]

orshot_batch_render

Read from the workspace's connected Google Sheets, Google Drive, Airtable, Shopify or HubSpot account to resolve the concrete IDs a workflow source needs. This is how you turn 'use my ABC spreadsheet' into a real spreadsheet `id` and `tab`: list the workspace's spreadsheets/Drive folders/Airtable bases+tables, then sample a source. A sample returns the field names plus up to 5 recent rows or records with their values, which can include names and emails from the user's own data, so the fields can be mapped. Call with no args for the provider/resource catalog. Always confirm the resolved name back to the user before creating a workflow.

orshot_get_workflow_connection_data

Configure the active workspace's Orshot Embed (white-label editor): activate it, set allowed domains (required for it to load anywhere), branding, feature toggles, events webhook, and custom buttons. Embed and some of its fields (events, webhook, downloads, custom buttons, asset picker, template-override protection) are not included in every plan; when the workspace's plan does not include one, the tool returns a plan-limit error and changes nothing. The signing secret cannot be read or written here. Omitted fields are left unchanged.

orshot_configure_embed

Finalize a direct upload started with orshot_create_upload_url: the kind is read from the key; the bytes are verified (magic bytes, size, completeness), processed exactly like the matching upload_brand_* tool, and saved to the workspace Brand Library. Returns the public direct_url (and for fonts the fontFamily to use in element styles).

orshot_confirm_upload

Legacy name for orshot_confirm_upload. Prefer orshot_confirm_upload — it is the primary tool and this name will be removed in a future release. Same inputs and output. Finalize a direct upload started with orshot_create_upload_url: the kind is read from the key; the bytes are verified (magic bytes, size, completeness), processed exactly like the matching upload_brand_* tool, and saved to the workspace Brand Library. Returns the public direct_url (and for fonts the fontFamily to use in element styles).

orshot_confirm_image_upload

Create a complete, reusable studio template for marketing, ads, social media, content creation, documents, or presentations. 📖 Read the **orshot://template-design-spec** resource first — it has the element structure, text rules, the verified animation schema, canvas sizes, and the common render gotchas (this tool also returns ⚠️ "Design checks" warnings; make one focused correction only when a warning is present). 🎨 For design quality, read the **orshot://design-spec/general** resource (the craft layer: hierarchy, 2-font typography, 60-30-10 color, brand-first, variety, anti-slop) — then the matching vertical pack, e.g. **orshot://design-pack/social-media** (formats, safe areas, archetypes, a concrete recipe). For recognised verticals this tool returns a "🎨 Design check" (layout-repetition, font/contrast, plus an informational brand-kit note). A clean check ends the authoring loop; correct only concrete findings. Brand assets are CONTEXTUAL — apply them only when the design should be on-brand (the prompt references the user's brand/product); a generic/standalone look needs no brand assets. DEFAULT: author the strongest complete design you can in this call — finished hierarchy and copy, production assets, reusable dynamic fields, intentional motion, and (for video) narration/ambient/captions/timing. Set includeThumbnails:true so creation doubles as visual QA. Do not deliberately submit a sparse scaffold. Make at most one focused update afterward, and only when the returned design checks or preview expose a real problem. Thumbnails are generated automatically in the background after creation — you do NOT need to render the template manually to preview it. They appear on the template (visible via orshot_get_studio_template) within a few seconds per page. The result returns updatedAt and pageMap: every page's page_id and every layer's element_id and parameter_id, including ids the server generated. Use them directly in a follow-up orshot_update_template_design or orshot_patch_template_elements call; do not call orshot_get_studio_template first. Recommended workflow: 1. In parallel: read the relevant design resources; call orshot_get_brand_kit for on-brand work; call orshot_get_audio_authoring_options for generated narration/SFX/captions. 2. orshot_create_template_design — create the complete template using the gathered context: - Use brand colors for backgrounds, text colors, accents - Use brand fonts (font name without extension) for fontFamily in text elements - Use brand images (direct_url) for logos, backgrounds, product shots - Use brand videos (direct_url) for video elements 3. If design checks or the thumbnail identify a concrete issue, make one targeted orshot_update_template_design call. Otherwise stop. ### Brand Asset Tag Matching When building templates, match brand assets by their tags to use the most relevant ones: - Tags like "logo", "icon" → use for logo placements - Tags like "background", "hero" → use for background images - Tags like "primary", "accent" → use for key color choices - Tags like "heading", "display" → use for title fonts - Tags like "body", "regular" → use for body text fonts IMPORTANT Element Format: - Position: use nested object { x: number, y: number } - Dimensions: use nested object { width: number, height: number } - Style values: px strings for lengths (fontSize: "40px"), unitless strings for fontWeight ("700") and minFontSize ("8"), NUMBER for lineHeight (1.2) Common canvas sizes: - Square: 1080×1080 (Instagram post) - Story: 1080×1920 (Instagram/TikTok story) - Slide: 1920×1080 (Presentations) - YouTube Thumbnail: 1280×720 - OG Image: 1200×630 - Twitter Header: 1500×500 ## Design Guidelines ### Canvas Properties Each page has a canvas object: { width, height, backgroundColor, backgroundImage, borderWidth, borderColor, borderStyle } - backgroundColor: Hex color or gradient, e.g. "#ffffff", "linear-gradient(180deg, #000 0%, #333 100%)" - backgroundImage: URL for a background image - borderWidth: Number in pixels (0 for no border) - borderColor: Color string, e.g. "rgba(0, 0, 0, 1)" - borderStyle: "solid" | "dashed" | "dotted" ### Element Structure Each element in pages_data[].elements[] needs: - id: Optional. Unique among the layers of its page; generated when omitted (returned in pageMap). Keep it when re-sending a design. - type: "text" | "image" | "shape" | "element" | "video" - position: { x: number, y: number } — position from top-left in pixels - dimensions: { width: number, height: number } — size in pixels - style: Object with CSS-like properties (length values as px strings; lineHeight as a NUMBER) - zIndex: Number for layer ordering (higher = on top) - For dynamic elements: parameterizable: true, parameterId: "snake_case_id", parameterType: "text" | "imageUrl" | "videoUrl" ### Common Element Properties (all types) - rotation: Number in degrees (0-360) - name: Optional display name for the layer - aspectRatioLocked: boolean (lock aspect ratio on resize) - clipPath: CSS clip-path value for masking, e.g. "polygon(...)", "circle(...)", "inset(...)" - style.opacity: 0-1 - style.mixBlendMode: "normal" | "multiply" | "screen" | "overlay" | "darken" | "lighten" | "color-dodge" | "color-burn" | "hard-light" | "soft-light" | "difference" | "exclusion" | "hue" | "saturation" | "color" | "luminosity" | "plus-lighter" - style.filter: CSS filter string, e.g. "blur(5px)", "grayscale(100%)", "brightness(1.2)" - style.dropShadowX: "0px" — drop shadow horizontal offset (separate from box shadow; applies as CSS filter) - style.dropShadowY: "0px" — drop shadow vertical offset - style.dropShadowBlur: "0px" — drop shadow blur radius - style.dropShadowColor: "rgba(0,0,0,0.2)" — drop shadow color ### Text Elements (type: "text") - content: Plain text string OR styled HTML for mixed formatting - Plain: "Hello World" - Styled (rich text): '<p><span style="color: red; font-weight: bold;">Bold Red</span> Normal</p>' - Multi-font styled text: '<p><span style="font-family: Playfair Display; font-size: 48px;">Elegant</span> <span style="font-family: Inter; font-size: 24px;">meets modern</span></p>' - Allowed tags: <b>, <strong>, <i>, <em>, <u>, <s>, <span>, <p>, <br> - Allowed CSS in style attr: color, font-family, font-size, font-weight, font-style, text-decoration, background-color, background-image, background-clip, letter-spacing, line-height, text-transform, padding - font-family in spans: Use any Google Font name. The renderer auto-loads fonts from Google Fonts CDN. - Use styled HTML when different words in the same text need different colors, sizes, fonts, or weights - style.textMode: "overflow" | "truncate" | "fit" — controls text overflow behavior: - "overflow" (default): text overflows the element boundaries - "truncate": text is cut off with ellipsis when it exceeds the element box - "fit": text auto-shrinks font size to fit within the element box (respects minFontSize) - style.minFontSize: String like "8" — minimum font size when textMode is "fit" - style.fontFamily: Google Font name — e.g. "Inter", "Prata", "DM Sans", "Playfair Display", "Sora", "Space Grotesk" - style.fontSize: "48px", "72px" (string with px) - style.fontWeight: "400", "700", "900" - style.fontStyle: "normal" | "italic" - style.color: Hex like "#FFFFFF" or gradient - style.textAlign: "left" | "center" | "right" | "justify" - style.verticalAlign: "flex-start" | "center" | "flex-end" — vertical text alignment (flex-start=top, center=middle, flex-end=bottom) - style.lineHeight: Number like 1.2, 1.4 - style.letterSpacing: "0px", "2px" - style.textTransform: "none" | "uppercase" | "lowercase" | "capitalize" - style.textDecoration: "none" | "underline" | "line-through" - style.paddingX: "6px", "10px" — horizontal padding inside the text box - style.paddingY: "6px", "10px" — vertical padding inside the text box - style.textStrokeWidth: "0px", "2px" — text outline/stroke width - style.textStrokeColor: "#000000" — text outline/stroke color - style.textBackgroundColor: Color value — background highlight behind text - style.textBackgroundRadius: "0px", "4px" — border radius of text background highlight - style.textBorderWidth: "0px", "2px" — border around the text element - style.textBorderColor: Color value — border color for text element - style.boxShadowX/Y/Blur/Color: Shadow properties Text sizing rules: - width: ~0.55 × fontSize × characterCount (tight fit, not full canvas) - height: fontSize × 1.3 for single line, multiply by lines × lineHeight for multi-line ### Shape Elements (type: "shape") - shapeType: "rectangle" | "circle" | "line" | "arrow" - style.fill: Color, rgba, or gradient (rectangle/circle) — the ONLY way a shape is painted. style.backgroundColor does NOTHING on a shape and leaves it opaque black. Gradient scrims over photos: a rectangle with fill "linear-gradient(180deg, rgba(0,0,0,0) 0%, rgba(0,0,0,0.9) 100%)". - style.stroke: Stroke color (rectangle/circle) - style.strokeWidth: "0px", "2px" (all shapes) - style.borderRadius: "0px", "12px", "50%" (rectangle only) - style.color: Color for line/arrow shapes - style.lineStyle: "solid" | "dashed" | "dotted" (line only) - style.opacity: 0-1 - style.boxShadowX/Y/Blur/Color: Shadow properties ### Image Elements (type: "image") - content: Publicly accessible image URL, OR a stock ref (see below) - STOCK: find refs with orshot_search_stock_media, then pass "stock://pixabay/photo/{id}" or "stock://unsplash/photo/{id}" as content and Orshot downloads, re-hosts and swaps in a permanent URL on save, recording the photographer credit on the element. Same for video layers: "stock://pixabay/video/{id}". Use this instead of hotlinking a provider CDN URL — those expire within about a day and the template breaks. An unresolvable ref fails the save rather than storing a dead link. - style.objectFit: "contain" | "cover" | "fill" - style.objectPosition: "left top" | "center top" | "right top" | "left center" | "center" | "right center" | "left bottom" | "center bottom" | "right bottom" - style.borderRadius, borderWidth, borderColor: Border properties - style.boxShadowX/Y/Blur/Color: Shadow properties - style.clipPath: CSS clip-path for masking ### SVG Elements (type: "element") - content: Inline data URI — "data:image/svg+xml,%3Csvg ..." - isSvg: true - style.svgColor: Hex color to tint - style.objectFit, objectPosition, borderRadius, borderWidth, borderColor, boxShadow*: Same as image ### Video Elements (type: "video") - content: Public mp4 URL - style.objectFit, objectPosition, borderRadius, borderWidth, borderColor, boxShadow*: Same as image - videoOptions: { loop: boolean, muted: boolean, duration: number, trim_start_time: number, trim_end_time: number } ### Waveform Elements (type: "waveform") - Visualize a page soundtrack: { type:"waveform", waveTrack:0, position, dimensions, style:{ fill:"#FFFFFF", barCount:48, barGap:3, borderRadius:"99px", waveAlign:"center"|"bottom", minBarHeight:8 } }. - waveTrack is the zero-based page.audioTracks index. Still previews use a deterministic shape; video renders animate from that track. No content field is needed. - Parameterize flat style keys for render-time changes such as "wave1.fill" or "wave1.barCount". ### Animations (transitions) Three INDEPENDENT phases per layer — enter, ambient, exit. Combine them; they are not alternatives. - transitions.enter: { type, duration, easing, reverse, customValues } - transitions.ambient: { type, duration, easing, customValues } — plays for the layer's WHOLE visible life, starting when enter finishes - transitions.exit: { type, duration, easing, reverse, customValues } - transitions.showAt: seconds when element appears - transitions.hideAt: seconds when exit starts (null = end). showAt+hideAt = per-layer visibility windows (the slideshow mechanism) Enter types: "none", "fadeIn", "slideInLeft", "slideInRight", "slideInUp", "slideInDown", "scaleIn", "scaleInUp", "scaleInDown", "rotateIn", "flipInX", "flipInY", "textReveal" (text only), "typewriter" (text only) Exit types: "none", "fadeOut", "slideOutLeft", "slideOutRight", "slideOutUp", "slideOutDown", "scaleOut", "scaleOutUp", "scaleOutDown", "rotateOut", "flipOutX", "flipOutY", "textRevealOut" Easing: "linear", "ease", "ease-in", "ease-out", "ease-in-out", "cubic-bezier(0.175, 0.885, 0.32, 1.275)" (bounce), "spring(stiffness, damping, velocity)" e.g. "spring(180, 12, 0)" (physics) customValues (intensity): opacity 0-1, offset ±500 (%), scale 0-5, angle (deg), granularity "word"|"char", overlap 10-90 Ambient types (mid-life motion): "none", "kenBurns" (SUSTAIN — one slow pan+zoom stretched across the layer's whole window, ignores duration), "pulse", "float", "drift", "wiggle", "shake", "slowRotate", "breathe" (all LOOP on their duration, default 4s) Ambient customValues: kenBurns zoom 0.5-2 (1.12) + panX/panY ±30%; pulse amount 1-1.5 (1.04); float offset 1-100px (12); drift offset 1-100px (10); wiggle angle 0.5-30deg (2); shake offset 1-40px (4); breathe opacity 0.1-1 (0.85) + scale 1-1.3 (1.02); slowRotate none Ambient easing: shake and slowRotate default to "linear" — a looping animation needs linear or the seam is visible. Everything else defaults to "ease-in-out". **Making video look designed, not generated.** The difference between a slideshow and something people watch: 1. EVERY full-bleed photo gets a kenBurns. A still image held for 3 seconds reads as a dead frame; a slow push makes it read as footage. Vary the direction between pages (panX +8 on one, -8 on the next) so consecutive scenes do not drift identically. 2. Ambient is SUBTLE by default and should stay there. kenBurns zoom 1.08-1.15, float offset 8-16, breathe near its defaults. If a viewer can describe the motion, it is too strong — except on a deliberate accent layer (a badge, a sticker) where wiggle or pulse is the joke. 3. Layer the phases: a headline can slideInUp (enter), breathe faintly (ambient), and fadeOut (exit). Ambient starts only once enter finishes, so they never fight. 4. Give ambient a SLOWER duration than the enter animation — 6-10s cycles against a 0.5s enter. Fast ambient reads as a glitch. 5. Pair with captions: videoOptions.captionPreset "karaoke" or "bold" plus subtitleStrokeWidth 2-3 keeps text readable over moving footage. 6. One pageTransition for the whole video, 0.3-0.5s. Different transitions per cut is the clearest tell of an auto-generated edit. ### Page audio + subtitles (video templates) - Call orshot_get_audio_authoring_options once for plan limits, models and voice ids. - page.audioTracks: max 10 mixed tracks. Existing audio: { url, trimStart, trimEnd, fadeIn, fadeOut, volume, offset, loop }. Narration: { tts:{ text, voiceId?, modelId?, parameterId? }, volume, ... }. Ambient/SFX: { sfx:{ prompt, seconds? }, volume:0.08-0.18, loop:true, fadeIn, fadeOut }. - Budget narration at about 2 words/second with 1 second of headroom. Create rejects likely clipping before it synthesizes or saves anything. - A strong social/marketing default is narration first, a restrained ambient bed second, and page.subtitle:{ source:"auto-audio", mode:"word-by-word", preset:"karaoke"|"bold", strokeWidth:2 }. - tts.parameterId makes SPOKEN COPY dynamic: single-page modifications { "voiceover":"New script" }, multi-page { "page1@voiceover":"..." }. - page.subtitle: { source:"auto-audio" (caption first track)|"upload", mode:"phrase"|"word-by-word", wordReveal:"appear"|"highlight"|"align", split:"word"|"letter", maxWordsPerLine OR maxCharsPerLine:1-80, styling keys }. maxCharsPerLine overrides the word budget; align re-centers the line as it grows. - page.videoDuration: default render duration (s) for this page. 🃏 PREVIEW CARDS — the user sees ONE final card, not your build steps. Every call of this tool drops a preview card into the chat; a stack of superseded cards reads as clutter and confuses the user. Create the complete design with includeThumbnails:true so that first card can be final. Stop when its preview and checks are clean. Only when they reveal a concrete issue, batch the correction into ONE focused update; the update card then replaces the create card as the result. Multiple cards are fine ONLY when the user asked to see variants or before/after.

orshot_create_template_design