orshot_add_brand_color
ChatGPTSave 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_analyze_media
ChatGPTMeasure a sound or a video's soundtrack (no AI; ffmpeg on the server): beats and tempo (bpm) to cut scenes and pop layers on the beat; silences with the keep ranges around them to tighten a talking-head clip; and, with jump_cut, the trimmed back-to-back video layers of that edit (alternate cuts punch in) to drop into pages_data. Takes a public http(s) URL of an audio or video file (a brand library audio/video, an upload, a stock track). Read-only: nothing is stored or charged. For music-reactive layers you do not need this: motion.reactive reads the page music at render time.
orshot_auth_status
ChatGPTCheck if authenticated and list available workspaces.
orshot_batch_render
ChatGPTLegacy 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_template
ChatGPTRender 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_configure_embed
ChatGPTConfigure 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_confirm_image_upload
ChatGPTLegacy 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_upload
ChatGPTFinalize 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_create_dynamic_url
ChatGPTMint (or update) a public, signed render URL for a studio template and return it ready to use. Use this when you need a STABLE, reusable, cacheable image link — OG/social images, email images, docs/dashboard embeds — rather than a one-off render (orshot_render_template). The returned URL renders the template on the fly; pass modification values as query params to produce variants from the same URL. Control access with active, maxRenders, validTill, and cacheForSeconds. Calling again for the same template updates only the fields you pass (partial update) and keeps the signature.
orshot_create_image_upload_url
ChatGPTLegacy name for orshot_create_upload_url. Prefer orshot_create_upload_url — it is the primary tool and this name will be removed in a future release. Same inputs and output. Get a short-lived signed URL to upload the raw BYTES of a file (image, font, video or audio) straight into Orshot storage — the way to ingest files larger than the ~7MB base64 ceiling, or to avoid pushing bytes through the conversation at all. When you run in a code sandbox that holds the user's attached file (for example at /mnt/user-data/uploads/), this is THE way to send it; never re-type a file as base64 into a tool argument. Flow: 1) call this with kind + contentType → 2) HTTP PUT the raw bytes to the returned uploadUrl (e.g. curl -X PUT --data-binary @clip.mp4 -H "Content-Type: video/mp4" "<uploadUrl>") → 3) call orshot_confirm_upload with the returned key. Confirm runs the same checks and processing as the matching upload_brand_ tool and saves the file to the workspace Brand Library. Limits: image 10MB, font 10MB, video 60MB, audio 50MB. The uploadUrl is on api.orshot.com, so a sandbox only needs `.orshot.com reachable. If the PUT returns 403 "Host not in allowlist", the sandbox has no network access: ask the user to turn on Settings → Capabilities → "Allow network egress" and add *.orshot.com` under Additional allowed domains, then start a new chat and retry with a fresh URL. URL expires in 5 minutes.
orshot_create_template_design
ChatGPTCreate 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)" - …
orshot_create_template_folder
ChatGPTCreate a studio template folder in the active workspace. Not available on every plan; when the workspace's plan does not include folders, the tool returns a plan-limit error and creates nothing. Names are unique per workspace and capped at 50 characters. Creating a folder does not move anything into it: follow up with orshot_move_template_to_folder for each template.
orshot_create_upload_url
ChatGPTGet a short-lived signed URL to upload the raw BYTES of a file (image, font, video or audio) straight into Orshot storage — the way to ingest files larger than the ~7MB base64 ceiling, or to avoid pushing bytes through the conversation at all. When you run in a code sandbox that holds the user's attached file (for example at /mnt/user-data/uploads/), this is THE way to send it; never re-type a file as base64 into a tool argument. Flow: 1) call this with kind + contentType → 2) HTTP PUT the raw bytes to the returned uploadUrl (e.g. curl -X PUT --data-binary @clip.mp4 -H "Content-Type: video/mp4" "<uploadUrl>") → 3) call orshot_confirm_upload with the returned key. Confirm runs the same checks and processing as the matching upload_brand_ tool and saves the file to the workspace Brand Library. Limits: image 10MB, font 10MB, video 60MB, audio 50MB. The uploadUrl is on api.orshot.com, so a sandbox only needs `.orshot.com reachable. If the PUT returns 403 "Host not in allowlist", the sandbox has no network access: ask the user to turn on Settings → Capabilities → "Allow network egress" and add *.orshot.com` under Additional allowed domains, then start a new chat and retry with a fresh URL. URL expires in 5 minutes.
orshot_create_workflow
ChatGPTCreate a workflow automation. Provide ordered steps (trigger → render → destination) built from orshot_list_workflow_nodes; validate first with orshot_validate_workflow. Defaults to 'draft'. Set status:'active' only when it's complete and runnable — an incomplete workflow can't be activated. Returns the created workflow plus any warnings.
orshot_delete_brand_audio
ChatGPTPermanently delete a brand audio file.
orshot_delete_brand_color
ChatGPTPermanently delete a saved brand color.
orshot_delete_brand_font
ChatGPTPermanently delete a custom font. Templates using this font will fall back to the default font.
orshot_delete_brand_image
ChatGPTPermanently delete a brand asset image.
orshot_delete_brand_video
ChatGPTPermanently delete a brand video.
orshot_delete_dynamic_url
ChatGPTDelete the dynamic URL for a template. The public URL stops working. Does not affect the template itself.
orshot_delete_template
ChatGPTPermanently delete a studio template. This action cannot be undone.
orshot_delete_template_folder
ChatGPTDelete a template folder. The folder itself is gone for good, but NO templates are deleted — every template inside returns to 'All templates' unfiled. Confirm with the user before calling this, since the grouping cannot be restored (re-filing each template is manual).
orshot_delete_workflow
ChatGPTDelete a workflow. This stops any future runs. Ask the user to confirm before deleting.
orshot_discard_social_post
ChatGPTPermanently delete a draft or scheduled post, in Orshot and on the platforms it was queued with. Use it to cancel something scheduled or clear a draft the user rejected. This cannot be undone, and it only works while the post is still held — an already-published post cannot be taken down from here.
orshot_duplicate_template
ChatGPTCreate a copy of an existing studio template. Returns the new template's ID.
orshot_generate_ai_audio
ChatGPTGenerate a standalone voiceover (text to speech) or ambient sound and save it to the workspace Brand Library. Returns a hosted MP3 URL you can place on any template audio track, use as the speech for orshot_generate_ai_video (pass the URL as audio), or hand to the user. Cost: AI credits, 1 per started 10 seconds (premium speech models 2). Identical text + voice + model is served from cache for free. Ambient sound is 1 to 22 seconds (loop it on a page for longer beds). Prefer template-level narration (pages_data[].audioTracks[].tts, see orshot_get_audio_authoring_options) when the audio belongs to a template render, because that stays editable and can be personalised per render. Use this tool for one-off tracks. Never generate in a loop; one track per request unless the user asks for another.
orshot_generate_ai_image
ChatGPTGenerate one image with AI from a prompt and up to 4 reference images, and save it to the workspace Brand Library. Returns a hosted PNG URL to use as an image element's content, or as imageRef for orshot_generate_ai_video. Use it for: - a product shot or scene stock does not have, with the user's product photo as a reference; - the first frame of an AI clip: the video models follow only their first frame, so to show the user's product in a clip, generate the still here (product photo as reference, the clip's aspect), then animate it with orshot_generate_ai_video imageRef. A photo attached in this chat goes in referenceAttachment (never invent a URL for it). The user's own product photo or logo that is not at a public URL yet (attached in the chat, or on their device): call orshot_upload_brand_image with no file and a short purpose to open the upload card, then use the direct_url it returns. A file in a terminal or sandbox: orshot_create_upload_url → PUT → orshot_confirm_upload. Pass that direct_url in references. Reach for free sources first: the user's own photos (orshot_get_brand_images), then orshot_search_stock_media. Cost: 2 AI credits per new image; the same inputs again return the cached image free. Not every workspace plan includes AI images: if the API answers ai-image-not-on-plan, tell the user plainly and use stock or the user's photos instead of retrying.
orshot_generate_ai_video
ChatGPTGenerate a short video clip with AI and save it to the workspace Brand Library. What you pass decides what happens: prompt only = footage from text; prompt + imageRef = animate that image; imageRef + lastImageRef = transition; imageRef + script (or audio) + consent = the person in the photo speaks. A photo attached in this chat goes in imageAttachment (never invent a URL for it). Use it for the clips a template needs — a background loop, a product hero shot, b-roll, a hook, a UGC-style product demo, a speaking portrait — then place the returned URL in a video element (orshot_create_template_design / orshot_patch_template_elements) and render. To generate per render instead, pass the same inputs on a render as "<videoParamId>.prompt", ".imageRef", ".preset", ".look", ".duration" modifications. Cost: 3 AI credits per second HD (5 s = 15), 6 per second Full HD. Charged only when a clip is delivered; the same inputs again return the cached clip for free. Speech needs a clear portrait; consent must come from the user. Not every workspace plan includes AI video: if the API answers ai-video-not-on-plan, tell the user plainly and fall back to orshot_search_stock_media or an existing clip from orshot_get_brand_videos instead of retrying. Timing: 5 s clips usually return inline (15-45 s). Longer clips and speech take 1-3 minutes — the tool then returns { pending: true, jobId } instead of waiting; collect the clip with orshot_get_render_job (id = jobId, waitSeconds: 25), calling again while finished is false. Never re-run this tool for a pending clip. Reuse before you generate: check orshot_get_brand_videos for an existing clip that fits, try orshot_search_stock_media for real footage (free, no AI credits), prefer animating a still you already have over text-only, and never regenerate in a loop — one clip per request unless the user asks for another take (fresh: true).
orshot_generate_image
ChatGPTLegacy name for orshot_render_template. Prefer orshot_render_template — it is the primary tool and this name will be removed in a future release. Same inputs and output. Render an image, PDF, or video from a studio template. For AI footage use orshot_generate_ai_video; AI images come from an image element's .prompt modification at render time. Supports rendering to multiple formats with optional quality, scale, page selection, and video options. For multi-page templates, use includePages to select specific pages (e.g. [1, 3]). For PDF output, use pdfDpi to control print quality (default 72, use 300 for print). For video/GIF output, use videoOptions to control duration, fps, and trimming — plus soundtrack (audioSource swaps/adds page audio tracks; use brand audio URLs), burned-in subtitles (subtitleSource auto-transcribes any voiceover URL via Whisper, or takes .srt/.vtt; style with the subtitle* keys, karaoke via subtitleMode "word-by-word"), global muting, and multi-page stitching (combinePages + pageTransition). CHECK LAYOUT WITH A STILL, NOT A VIDEO. Before spending an mp4 (60-90s + video credits), verify composition with format:"png", includePages:[N], responseType:"base64" — about 1s, one image credit, and you SEE the frame. Layout, contrast, scrims, safe-area and text fit are all identical to the video's first frame; only motion is missing. Fix what you see, then render the video ONCE. Rendering an mp4 to find out whether a layer is in the right place is the most expensive mistake available here. LONG RENDERS: video/subtitle renders can take 60-90s. This server streams progress to keep the request alive; if your client times out anyway, the render still completes server-side (credits are used) — recover the output URL via orshot_list_workspace_logs instead of re-rendering. For guaranteed-no-timeout operation (or several long renders in a row), pass background: true — the tool returns a jobId immediately; collect it with orshot_get_render_job using waitSeconds: 25 (the server holds and polls for you; call again while finished is false). SMART RESIZE (one design → any size): the design is deterministically re-laid-out to fit a new canvas. - Use size to render the design at a DIFFERENT size (e.g. size: "instagram-story" or "1080x1080" or { width, height }). - Use extraSizes to ADD extra sized copies in the SAME call (e.g. extraSizes: ["instagram-story","1080x1080","1200x630"]) — the response nests an "extraSizes" array of { name, size, width, height, url } beside each page. Perfect when a user asks for one design across multiple social formats. RESPONSE SHAPE (so you surface the right links to the user): - Single-page: data.data is one object → data.data.content is the main output URL; data.data.extraSizes (if requested) is an array of variants, each { name, size, width, height, content } where content is that variant's URL. - Multi-page: data.data is an array (one per page); each page has its own content and extraSizes. This tool also prepends a ready-to-share summary of every output + variant URL, so present those links to the user; a per-variant error means that one size failed (the others still succeeded). Optionally publish the rendered output to connected social accounts by including the "publish" parameter. IMPORTANT: Only include the "publish" parameter when the user explicitly asks to publish or post to social media. Never auto-publish. Use orshot_list_social_accounts first to get the account IDs needed for publish.accounts. 🃏 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_generate_image_from_library
ChatGPTLegacy name for orshot_render_library_template. Prefer orshot_render_library_template — it is the primary tool and this name will be removed in a future release. Same inputs and output. Render an image from a pre-built Orshot library template. Use orshot_list_library_templates to see available templates and orshot_get_library_template_modifications to see what parameters each accepts. Common templates: - website-screenshot: Capture a webpage (modifications: { websiteUrl, fullCapture?, delay?, width?, height? }) - tweet-image: Screenshot a tweet (modifications: { tweetUrl }) - beautify-screenshot: Add padding/shadow to a screenshot (modifications: { screenshotImage }) - open-graph-image-1: OG image for blog posts (modifications: { title, description }) - image-background-removal-api: Remove image background (modifications: { inputImage }) - dynamic-image-resizer: Resize/crop an image (modifications: { imageUrl, width?, height?, resizeType? }) 🃏 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_get_account
ChatGPTFetch the authenticated user's account: profile (name, email), workspaces, current plan, and live usage — render_credits (template renders: 1 image, PDF page or video second = 1 credit) and ai_credits (one meter shared by AI images, AI video and voiceovers, with each capability's caps and rates nested inside it), member/workspace/rate limits, and which features (workflows, embed, video rendering) the plan includes. Call this to check whether a feature is available on the current plan. For the voice catalog and complete audio authoring limits, use orshot_get_audio_authoring_options.
orshot_get_ai_video_presets
ChatGPTList 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_audio_authoring_options
ChatGPTCall once before creating or updating a template with generated narration, ambient audio, sound effects, or automatic captions. Returns the account's AI credit position (voice shares the pool with AI images and video) together with every supported speech model and the public voice catalog, so an agent can choose valid settings and author the complete video template without follow-up discovery calls. Usage is account-wide across workspaces; the active workspace is retained only for attribution. This tool does NOT generate audio. For a standalone track use orshot_generate_ai_audio. For template narration put it in pages_data[].audioTracks[].tts and ambient/SFX in pages_data[].audioTracks[].sfx when calling orshot_create_template_design or orshot_update_template_design. Saving a new recipe synthesizes a hosted track and spends AI credits (1 per started 10 s, premium models 2); an identical cached recipe is free. For dynamic personalized narration, set tts.parameterId and pass new spoken text through render modifications. Author narration at roughly 2 words per second with one second of headroom; create, update, patch, and dynamic video render tools preflight likely clipping before synthesis.
orshot_get_best_time_to_post
ChatGPTWhen this workspace's audience actually engages, computed from its own posting history: the best day-and-hour slots (UTC), how posting frequency relates to engagement per platform, and how fast engagement decays after publishing. Use it before scheduling to pick a time, and prefer it over generic best-time advice because it is derived from these accounts. Slots need posting history to appear, so a new workspace can legitimately come back empty. Read-only, shares the 10 requests per minute analytics budget.
orshot_get_brand_audio
ChatGPTList audio files in the workspace's brand library. Use tag filtering to find tracks for specific purposes (e.g. "music", "voiceover", "sfx"). Tags use AND logic. Returns up to 100 audio files with metadata (duration, format, file size). Use a track's direct_url as the audio source on video template pages.
orshot_get_brand_colors
ChatGPTList saved brand colors in the workspace. Supports hex, rgb, hsl, and gradient types. Use tag filtering to find colors for specific purposes (e.g. "primary", "background", "accent"). Tags use AND logic. Returns up to 200 colors.
orshot_get_brand_fonts
ChatGPTList custom fonts uploaded to the workspace. Use tag filtering to find fonts for specific purposes (e.g. "heading", "body", "brand"). Tags use AND logic. Returns up to 100 fonts. When creating templates, use the font's name field (without extension) as the fontFamily in element styles. Custom fonts take priority over Google Fonts.
orshot_get_brand_images
ChatGPTList brand asset images in the workspace. Use tag filtering to find assets for specific purposes (e.g. "logo", "background", "product"). Tags use AND logic — all specified tags must match. Returns up to 100 images, newest first.
orshot_get_brand_kit
ChatGPTFetch all brand assets at once — images, colors, fonts, videos, and audio. Optionally filter each type by tags. This is the recommended starting point when creating branded templates. Use tag filtering to get only relevant assets for your design task. Example workflow: 1. orshot_get_brand_kit with tags matching your design purpose (e.g. "social-media") 2. Use returned brand colors for backgrounds, text colors 3. Use returned brand fonts for text elements (set fontFamily to the font name without extension) 4. Use returned brand images for logos, backgrounds, product shots 5. Use returned brand videos for video elements 6. Use returned brand audio for background music / voiceover on video templates
orshot_get_brand_videos
ChatGPTList brand videos in the workspace. Use tag filtering to find videos for specific purposes (e.g. "intro", "background", "product"). Tags use AND logic. Returns up to 100 videos with metadata (width, height, duration, format).
orshot_get_design_spec
ChatGPTRead the rules and references a template must follow. Call with topic "template-json" once per session before the first orshot_create_template_design and whenever a create or update returns a design check or schema error; call the other topics when you build that kind of thing. Topics: "template-json" (element structure, text rules, the animation schema, canvas sizes and render gotchas; read once per session before the first create, and when a design check fails); "craft" (the craft layer every template follows: hierarchy, two fonts, 60-30-10 colour, composition, anti-slop); "social-pack" (social posts, stories and carousels: formats and safe areas, what makes a post look designed, drawn graphics, a concrete element tree); "video-pack" (any template with motion or video: the first second, scenes, eases, holds, words on screen, banned defaults); "film" (the film schema, exact: the film block, page.transitionIn, keyframes, tracks and clips, limits); "film-pack" (film craft: beat sheet, cuts by cause, continuity cuts, music, scene recipes, checks); "layout" (the layout solver grammar: page.layout rows, stacks, grids, rings, anchors, safe areas, snippets); "formats-catalogue" (every format by name and id (section = post, ad, video or thumbnail to filter)); "post-formats" (the whole post formats reference (section = a format id or heading words for one format)); "ad-formats" (the whole ad formats reference (section = a format id or heading words for one format)); "video-post-formats" (the whole video post formats reference (section = a format id or heading words for one format)); "youtube-thumbnail-layouts" (the whole thumbnail layouts reference (section = a layout number or name)). Optional section returns one heading of the topic instead of all of it. orshot_plan_design already returns the matched formats, so read a whole formats reference only to browse.
orshot_get_dynamic_url
ChatGPTReturn the dynamic URL configuration (signed URL, active state, limits, expiry, view count) for a template, or a not-found error if none is configured.
orshot_get_embed_config
ChatGPTFetch the active workspace's Orshot Embed (white-label editor) configuration: whether it's set up, allowed domains, branding (title, icon, accent color), feature toggles, webhook, custom buttons, view count, and — importantly — whether the CURRENT PLAN allows embed and which embed features it includes. The signing secret is never returned through the API; the response includes the dashboard URL where the user can view it.
orshot_get_library_template_modifications
ChatGPTGet the available modification parameters for a specific library template. Use this to discover what fields you can customize (e.g. websiteUrl for website-screenshot, tweetUrl for tweet-image).
orshot_get_render_job
ChatGPTCheck the status of an async render job (video, PDF, or image) started in the background. Returns status (queued | processing | succeeded | failed | canceled), finished (true once terminal), and on success the output URLs. Pass waitSeconds: 25 and the SERVER holds the call, checking the job until it finishes or the wait elapses — then simply call again while finished is false (typically 1-6 calls for a video). Do not try to wait between calls yourself; agents have no sleep primitive, which is exactly what waitSeconds replaces. Job records expire 7 days after creation; the rendered file itself is unaffected by job expiry. Do NOT re-render while a job is queued or processing — that double-spends credits.
orshot_get_social_analytics
ChatGPTHow this workspace's social posts performed: impressions, engagement, clicks and engagement rate per post, plus follower counts per account. Start here for "how are we doing", "what worked" or "which post did best" — sort by engagement to rank winners. Covers posts synced from the platforms too, not just ones published through Orshot, so there is history from day one. Results are paginated: the reply states the workspace total and whether more pages exist, so quote the stated total rather than counting the rows you can see. Read-only. Analytics allows 10 requests per minute per workspace and responses are cached, so ask once and reuse the answer rather than polling.
orshot_get_social_analytics_daily
ChatGPTDay-by-day social metrics for this workspace, plus a per-platform breakdown of the whole window. Use it for trends over time ("are impressions going up?", "how did last week compare"), not for individual posts — orshot_get_social_analytics covers those. Read-only, and shares the 10 requests per minute analytics budget.
orshot_get_social_followers
ChatGPTFollower counts and growth for each connected social account, with a day-by-day follower history. Use it for "how is our audience growing", "which account grew this month" or to compare accounts. For post performance use orshot_get_social_analytics instead. Read-only, shares the 10 requests per minute analytics budget.
orshot_get_social_post
ChatGPTRead one post in full: caption, media, and per-account status with live post URLs or failure reasons. This is how you confirm what actually happened to a post — including confirming that a draft is still held and has NOT been published. Posts still in flight are refreshed from the platforms first, so there is no separate sync step. In clients that support inline previews this also renders the post as a visual card (media + caption + status), so call this when the user asks to see/show a post.
orshot_get_social_post_analytics
ChatGPTEverything measured for a single post: impressions, reach, likes, comments, shares, saves, clicks, engagement rate, per-platform results, and a day-by-day timeline since it published. Use it after orshot_get_social_analytics identifies a post worth understanding. A post published in the last few minutes will show no numbers yet — that is normal, not an error, so do not retry in a loop. Read-only, shares the 10 requests per minute analytics budget.
orshot_get_studio_template
ChatGPTGet full details of a studio template including design structure, every page_id, element_id and parameterId, and updated_at (the base_updated_at for a write). (An API-key connection gets page ids and dynamic parameters only; pageMapPartial is then true.) One call is enough before editing a template you have not written to in this conversation; create and update results already return the same ids (pageMap) and updatedAt. Shows the template's stored page thumbnails as an inline preview card (multipage templates get a swipeable carousel) — use this when the user asks what a template looks like. 🃏 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_get_studio_template_modifications
ChatGPTGet the available modification parameters for a studio template. Returns the list of dynamic fields (text, image, etc.) that can be customized when rendering via orshot_render_template. The response contains a "modifications" array where each entry has: - id: The key to use in the modifications object when rendering. For single-page templates this is just the parameterId (e.g. "headline"). For multi-page templates it is prefixed with the page number (e.g. "page1@headline", "page2@cta_text"). - type: The parameter type — "text", "imageUrl", "videoUrl", "backgroundColor", "fill", "color" - description: Human-readable label Beyond these listed keys, EVERY parameterizable element also accepts implicit dot-notation render props (not listed in the response): - Style: "paramId.fontSize": "64px", "paramId.color": "#FF0000", etc. - Animation timing (video renders): "paramId.showAt", "paramId.hideAt", "paramId.enterType"/"enterDuration"/"enterEasing"/"enterReverse", "paramId.exitType"/"exitDuration"/"exitEasing"/"exitReverse", "paramId.ambientType"/"ambientDuration"/"ambientEasing" - Video elements: "paramId.trimStart", "paramId.trimEnd", "paramId.muted", "paramId.loop" - Generated narration: a track with tts.parameterId is listed here with type "voiceText"; pass new spoken copy under that id. Existing soundtrack-file swaps use videoOptions.audioSource. There is no generic audio parameterType. - element: The element details (type, current default content, parameterId) Also returns "modifications_json" — a flat key→defaultValue map you can use as a starting point and override specific keys.
orshot_get_template_sharing
ChatGPTGet a studio template's public sharing state — whether it's shared, the public link, the embed page + ready-to-paste iframe code, whether visitors can duplicate it, and view/copy stats. Use orshot_update_template_sharing to change any of it.
orshot_get_workflow
ChatGPTGet a workflow's full definition — status, trigger, and its ordered steps (with config). In clients that support inline previews this also renders the workflow as a visual card (node pipeline with icons), so ALWAYS call this when the user asks to see/show a workflow — never build your own visualization of it.
orshot_get_workflow_connection_data
ChatGPTRead 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_run
ChatGPTGet one run's detail including per-node results — which step ran, its status, output, and any error. Use this to debug why a workflow run failed.
orshot_get_workflow_sharing
ChatGPTGet a workflow's public sharing state — whether it's shared, the public link and embed page, and whether its template is included when someone copies it. Use orshot_update_workflow_sharing to change any of it.
orshot_get_workflow_webhook
ChatGPTGet (or create) the inbound webhook URL for a webhook-triggered workflow — the URL to POST events to from your own app or a tool like Zapier/Make, plus whether a signing secret is set (the secret itself is shown only in the Orshot dashboard). Only relevant for workflows whose trigger is a webhook.
orshot_get_workspace
ChatGPTFetch the active workspace's details: title, slug, dashboard URL, creation date, and any saved onboarding answers (goals, cadence, destinations recorded by orshot_update_workspace). Use the onboarding block to pick up where a previous setup session left off instead of re-asking the user.
orshot_import_brand_website
ChatGPTSave logos, colours and fonts from a website into the workspace Brand Library, so later renders can use them by name or tag. Two ways to call it. Preferred: run orshot_scan_brand_website first, show the user what was found, then pass its scanId plus the images/colors/fonts keys they chose. Or pass url alone to scan and import in one step. With no keys and no min_confidence, it takes only high-confidence assets (0.7 and above) — the ones the scan is sure belong to that site. Lower the bar to take more, raise it to take less. Anything below the bar, already in the library, or no longer reachable is reported in skipped rather than failing the call. Every import returns an importId: pass it to orshot_undo_brand_import to remove everything it wrote.
orshot_list_connected_integrations
ChatGPTList the third-party integrations (Google Sheets, Drive, Airtable, Shopify, HubSpot, Slack, Dropbox, S3, WordPress) this workspace has connected, with the account each is connected as. Use before building a workflow to know which sources/destinations are wired (never returns tokens).
orshot_list_library_templates
ChatGPTList all available pre-built library templates (website-screenshot, tweet-image, OG images, mockups, etc.).
orshot_list_social_accounts
ChatGPTList all connected social media accounts in the active workspace. Returns each account's id (numeric), platform, account_name, account_username, and account_avatar. Use the numeric id values in the "publish.accounts" array of orshot_render_template to post renders directly to social media. Supported platforms: Twitter/X, Instagram, LinkedIn, Pinterest, Facebook, TikTok, YouTube, Threads, Bluesky, Reddit, Telegram, Snapchat, Google Business.
orshot_list_social_posts
ChatGPTList this workspace's social posts, newest first — drafts, scheduled, published and failed. Use it to find a post's id before reading, revising, sending or discarding it, to show the user what is waiting to go out (status: "draft"), or to check what recently published. Paginated with an opaque cursor. This is a text listing only — when the user asks to SEE or SHOW a post, follow up with orshot_get_social_post: it renders the post as an inline visual card.
orshot_list_studio_templates
ChatGPTList studio templates in the active workspace. Supports server-side search by name and filtering by tags. E2E/test/delete-me fixtures are hidden by default so creative discovery is not polluted; set includeTestTemplates only when debugging them. This is a text listing only — when the user asks to SEE or SHOW a template, follow up with orshot_get_studio_template: it renders the template as an inline visual card.
orshot_list_template_folders
ChatGPTList the studio template folders in the active workspace, in the same order the dashboard sidebar shows them. Start here before creating a folder (names are unique per workspace) or before moving templates, since orshot_move_template_to_folder needs a folder id.
orshot_list_workflow_nodes
ChatGPTDiscover the building blocks for workflow automations — every trigger, source, render, and destination node, with its config schema and whether the current plan can use it. Call this before building a workflow. Pass key to get one node's full config schema.
orshot_list_workflow_runs
ChatGPTList a workflow's recent runs with their status and timing.
orshot_list_workflows
ChatGPTList the workspace's workflow automations, newest first. Optionally filter by status. This is a text listing only — when the user asks to SEE or SHOW a workflow, follow up with orshot_get_workflow on the one they mean: it renders an inline visual card of the node pipeline.
orshot_list_workspace_logs
ChatGPTList render/API logs for the active workspace, newest first. Each entry is a trimmed summary: status, source, template, output format, render time, credits used, and the OUTPUT URL (main render; extra-size variants noted). Results are paginated with an opaque cursor — pass the returned next_cursor as "cursor" to fetch the next (older) page. Also the RECOVERY path for renders whose tool call timed out client-side: a completed render's output URL is here even when the render response never made it back — filter by template_id and/or format: "mp4" and take the newest entry instead of re-rendering (which would double-spend credits). Useful for reviewing recent render activity, debugging failed renders (filter status: "error"), or auditing credit usage over a date range.
orshot_list_workspaces
ChatGPTList all workspaces you have access to via OAuth.
orshot_move_template_to_folder
ChatGPTFile a studio template into a folder, or pull it out of whatever folder it is in. Pass folderId to file it; pass folderId: null to unfile it back to 'All templates'. A template lives in at most one folder, so moving it into a new folder removes it from the old one. The folder must belong to the active workspace.
orshot_patch_template_elements
ChatGPTUpdate individual element values in a saved template without re-sending the whole design. Use this to change text content, swap images, update colors, or modify style properties on specific elements by their parameterId. Use it for targeted changes; use orshot_update_template_design for a full redesign. Two modes (send exactly one): - data: new values for existing layers (below). Each item names its page by "page" (1-based number) or "page_id" (stable; keeps pointing at the same page after pages are added, deleted or reordered). Add "elementId" to an item to target any layer, dynamic or not, by its element id; its keys then use "element" as the base: { page: 1, elementId: "el-9", modifications: { "element": "New text", "element.fontSize": "64px" } }. - layer_operations: structural layer edits, applied in order, all or nothing: add, duplicate, replace, update, delete or move (restack) a layer, including SAVED animation timing and motion: [{ op: "update", parameter_id: "headline", page: 1, set: { transitions: { enter: { type: "slideInUp", duration: 0.6 }, showAt: 1.2 } } }], [{ op: "add", page: 2, element: { ...layer }, position: { above: "bg" } }], [{ op: "delete", element_id: "sticker-3" }]. Operations name a page by page_id or page too. Page-level edits (add/duplicate/delete/reorder pages) → orshot_update_template_design page_operations. Supports dot notation: "headline.fontSize" (style), "headline.x" / "headline.width" (position and size), "photo.crop" (object), "photo.alt". Ids and row version: use the pageMap (each page's page_id, each layer's element_id and parameter_id) from your last orshot_create_template_design / orshot_update_template_design result, or from orshot_get_studio_template for a template you have not seen in this conversation. Patching never changes existing ids: data returns the same pageMap, and layer_operations report the ids of layers they add in operationResults. base_updated_at is optional in both modes: pass the last updatedAt to refuse the write (409) if someone else changed the template since; a call that sends it gets the new updatedAt back for the next write. Examples: Single element: { data: [{ page: 1, modifications: { "headline": "New Title" } }] } By page id: { data: [{ page_id: "b2c3d4e5-…", modifications: { "headline": "New Title" } }] } Multiple: { data: [{ page: 1, modifications: { "headline": "Title", "hero_image": "https://..." } }] } Dynamic narration: { data: [{ page: 1, modifications: { "voiceover": "A shorter script that fits the page." } }] } Style update: { data: [{ page: 1, modifications: { "headline.fontSize": "64px", "headline.color": "#FF0000" } }] } Move/resize: { data: [{ page: 1, modifications: { "headline.x": 64, "headline.y": 120, "headline.width": 800 } }] } Crop an image: { data: [{ page: 1, modifications: { "photo.crop": { "x": 10, "y": 10, "width": 60, "height": 60 } } }] } Multi-page: { data: [{ page: 1, modifications: { "title": "Page 1" } }, { page: 2, modifications: { "title": "Page 2" } }] } WHAT LANDS (max 50 mods/request): content (text, image/video URLs incl. data: URIs), generated narration fields advertised as voiceText, style props, position (x, y) and size (width, height), crop objects, and .alt. A "pageN@key" prefix inside an item retargets that key to page N. Voice text is duration-checked before synthesis. Replacing a video clears stale source-duration metadata before save. ⚠️ NOT patchable — these are render-time only and come back in the response's "skipped" array, so nothing is stored: - Animation/transition keys ("photo6.showAt", hideAt/enterType/...) and video dynamics (.trimStart/.muted/.loop) in data. For SAVED timing use layer_operations ({ op: "update", set: { transitions } }); for ONE render pass them as modifications on orshot_render_template ("photo6.showAt": 2.1 works there). - AI prompt keys (.prompt/.promptMode/.promptReferences), .href and .contentType. - Page-level props (canvas, audioTracks, subtitle), except the first-class voiceText fields returned by orshot_get_studio_template_modifications. The response separates outcomes: "…
orshot_place_image
ChatGPTAdd an image element to an existing template in ONE call: fetches the current design, appends the image on the chosen page, and saves. Use a public URL or a brand-asset direct_url (upload first via orshot_upload_brand_image or the upload-url flow). ⚠️ Reads the current design and writes it back in full. If someone is editing this template in the Studio editor at the same moment, one set of changes can overwrite the other (last write wins) — avoid running while the template is open in an editor.
orshot_plan_design
ChatGPTPlan a new design before creating it (orshot_create_template_design) when the user has not fixed the look. It computes a complete design plan from the context you pass: a palette generated from the subject's own colours (or the brand kit) with contrast checked, a font pairing chosen by tone with its type scale in px for this canvas, a layout with slot rectangles in px on a margin grid, the material and devices, and for carousels an inside-slide system. Ground structures follow how real template catalogues vary per category, and the result rotates away from what this workspace made recently. Fill the context well, it drives everything: subject_colours = 2-4 hex colours of real things in the subject's world (coffee: espresso, crema, roast orange; sea: deep water, foam, sand), or the colors of the stock photo you will use (orshot_search_stock_media returns them per result; search first when the photo leads); tone axes; headline_words and items. Build the direction it returns, exactly (ask for count 2-4 only when the user wants options to compare, then use the one they pick): its colours are the only hues besides photo colours, its fonts the only fonts, its slots the layout. Not for edits to an existing template.
orshot_preview_element
ChatGPTPreview a design element before it goes into a template: an icon, logo, illustration or any drawn SVG is rendered to a PNG by the same engine that renders templates (fitted into width × height, objectFit contain) and returned as an image you can look at. Free: no credits, nothing stored. Each item is the layer as you would write it in pages_data: type: "element" with the drawing as svg_markup (or content as a data:image/svg+xml URI). Use it while drawing (orshot://film-spec §3 "Drawn vectors"): draw, preview, correct the shapes and colours against the brief or the reference, then put the same markup on the layer. Up to 6 elements per call; a drawing that animates (motion.svg) previews at its REST pose, which is how it should be drawn. SVG elements are supported today; another type returns an error naming the supported types.
orshot_publish_social_draft
ChatGPTPublish a draft or scheduled post NOW, to the accounts it already targets. This is the irreversible step: it puts the post in front of a real audience and it cannot be recalled through Orshot. Only call it when the user has explicitly approved this specific post. If they only asked you to prepare something, stop at the draft. To change what goes out, revise it first with orshot_update_social_post. To send to different accounts, revise the accounts first.
orshot_publish_social_post
ChatGPTPost media the workspace ALREADY has to connected social accounts — a previous render, a brand image, or any public URL. Use this instead of orshot_render_template's publish parameter whenever you are not rendering something new in the same breath. orshot_render_template can only attach the image it just produced; this takes any media URL. ONLY call this when the user explicitly asks to post, draft or schedule. It writes to accounts the public can see. Default to status: "draft" when the user says draft, "let me look first", "don't post yet", or anything else short of an explicit go-ahead. A draft is held by Orshot and reaches nobody until orshot_publish_social_draft sends it. Workflow: orshot_list_social_accounts (get IDs) → confirm the targets with the user → call this. Format rules: PDFs can't be posted anywhere; video can't go to Google Business; images can't go to TikTok or YouTube. Incompatible accounts come back as failed with a reason rather than sinking the whole post.
orshot_remove_background
ChatGPTRemove the background from an image and get a transparent PNG back as a permanent URL, ready to place as an image layer (objectFit "contain"): a product or person breaking out of a colour block, a founder in front of giant type, a logo lifted off its white box, a sticker cut-out. Works on any public image URL: a brand library image, an upload, a render, a product photo. Spends the workspace's AI credits on every call; for a stock result use orshot_use_stock_media with cutout: true instead, which reuses earlier cut-outs for free. Best on photos and flat illustrations with one clear subject.
orshot_rename_template_folder
ChatGPTRename an existing template folder. The templates inside are untouched. Same name rules as creation: 1-50 characters, unique within the workspace.
orshot_render_library_template
ChatGPTRender an image from a pre-built Orshot library template. Use orshot_list_library_templates to see available templates and orshot_get_library_template_modifications to see what parameters each accepts. Common templates: - website-screenshot: Capture a webpage (modifications: { websiteUrl, fullCapture?, delay?, width?, height? }) - tweet-image: Screenshot a tweet (modifications: { tweetUrl }) - beautify-screenshot: Add padding/shadow to a screenshot (modifications: { screenshotImage }) - open-graph-image-1: OG image for blog posts (modifications: { title, description }) - image-background-removal-api: Remove image background (modifications: { inputImage }) - dynamic-image-resizer: Resize/crop an image (modifications: { imageUrl, width?, height?, resizeType? }) 🃏 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_render_template
ChatGPTRender an image, PDF, or video from a studio template. For AI footage use orshot_generate_ai_video; AI images come from an image element's .prompt modification at render time. Supports rendering to multiple formats with optional quality, scale, page selection, and video options. For multi-page templates, use includePages to select specific pages (e.g. [1, 3]). For PDF output, use pdfDpi to control print quality (default 72, use 300 for print). For video/GIF output, use videoOptions to control duration, fps, and trimming — plus soundtrack (audioSource swaps/adds page audio tracks; use brand audio URLs), burned-in subtitles (subtitleSource auto-transcribes any voiceover URL via Whisper, or takes .srt/.vtt; style with the subtitle* keys, karaoke via subtitleMode "word-by-word"), global muting, and multi-page stitching (combinePages + pageTransition). CHECK LAYOUT WITH A STILL, NOT A VIDEO. Before spending an mp4 (60-90s + video credits), verify composition with format:"png", includePages:[N], responseType:"base64" — about 1s, one image credit, and you SEE the frame. Layout, contrast, scrims, safe-area and text fit are all identical to the video's first frame; only motion is missing. Fix what you see, then render the video ONCE. Rendering an mp4 to find out whether a layer is in the right place is the most expensive mistake available here. LONG RENDERS: video/subtitle renders can take 60-90s. This server streams progress to keep the request alive; if your client times out anyway, the render still completes server-side (credits are used) — recover the output URL via orshot_list_workspace_logs instead of re-rendering. For guaranteed-no-timeout operation (or several long renders in a row), pass background: true — the tool returns a jobId immediately; collect it with orshot_get_render_job using waitSeconds: 25 (the server holds and polls for you; call again while finished is false). SMART RESIZE (one design → any size): the design is deterministically re-laid-out to fit a new canvas. - Use size to render the design at a DIFFERENT size (e.g. size: "instagram-story" or "1080x1080" or { width, height }). - Use extraSizes to ADD extra sized copies in the SAME call (e.g. extraSizes: ["instagram-story","1080x1080","1200x630"]) — the response nests an "extraSizes" array of { name, size, width, height, url } beside each page. Perfect when a user asks for one design across multiple social formats. RESPONSE SHAPE (so you surface the right links to the user): - Single-page: data.data is one object → data.data.content is the main output URL; data.data.extraSizes (if requested) is an array of variants, each { name, size, width, height, content } where content is that variant's URL. - Multi-page: data.data is an array (one per page); each page has its own content and extraSizes. This tool also prepends a ready-to-share summary of every output + variant URL, so present those links to the user; a per-variant error means that one size failed (the others still succeeded). Optionally publish the rendered output to connected social accounts by including the "publish" parameter. IMPORTANT: Only include the "publish" parameter when the user explicitly asks to publish or post to social media. Never auto-publish. Use orshot_list_social_accounts first to get the account IDs needed for publish.accounts. 🃏 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_reset_dynamic_url_signature
ChatGPTRotate the signature for a template's dynamic URL. Every previously shared URL for this template stops working immediately — use it to revoke access. Irreversible: confirm with the user before calling. Returns the new signed URL.
orshot_run_workflow
ChatGPTManually trigger a workflow run now (in addition to its normal trigger). Returns the queued run id; use orshot_get_workflow_run to watch it. The workflow must be complete/runnable.
orshot_scan_brand_website
ChatGPTRead a public website and report the logos, colours and fonts it uses, WITHOUT saving anything. Use this first when someone says "set up my brand from acme.com" or "make this on-brand" and the Brand Library is empty. Every row comes back with a confidence from 0 to 1 and the reasons behind it. Above ~0.8 is almost always the real thing; below ~0.5 is usually a customer logo, a press badge or an incidental colour. Show the user what was found, then call orshot_import_brand_website with the keys they pick, or with min_confidence to take everything above a bar. Reads the page, its stylesheets, web manifest and font files, and renders it once to confirm what actually paints. Sites that block automated readers come back with a blocked note and just their public icon.
orshot_search_stock_media
ChatGPTSearch free stock photos and videos (Pixabay, Unsplash) the way the Studio's stock picker does. Returns 24 results per page by default (per_page up to 50), each with a stable ref (stock://provider/type/id), size, duration and the photographer credit. Use a result in one of two ways: - put its ref directly as an image or video element's content in orshot_create_template_design / orshot_update_template_design / orshot_patch_template_elements — Orshot downloads and re-hosts it on save and keeps the credit on the element; - or call orshot_use_stock_media with the ref to get a permanent URL now (also saved to the Brand Library). Never place previewUrl in a template: it is a temporary provider link. Prefer stock over AI generation when real footage or photography of the subject exists; it costs no AI credits. Licence: free for commercial use, no attribution required on rendered output.
orshot_set_active_workspace
ChatGPTSet which workspace to use for subsequent operations.
orshot_suggest_workflows
ChatGPTGiven a studio template, returns the 2-3 ready-made automations that fit it best (matched on the template's content, shape, and which integrations the workspace already has connected). Each suggestion comes back DRAFT-READY: its steps already have this template wired into the render step — pass them straight to orshot_create_workflow (status 'draft') and share the returned edit link. Call this right after rendering or designing a template to pitch a specific automation instead of a generic one.
orshot_undo_brand_import
ChatGPTRemove everything one brand-website import wrote — the images, colours and fonts it added, and their stored files. Takes the importId returned by orshot_import_brand_website. Available for 7 days after the import; it never touches assets that import did not create.
orshot_update_brand_audio_tags
ChatGPTReplace tags on a brand audio file. Pass empty array to clear all tags.
orshot_update_brand_color_tags
ChatGPTReplace tags on a saved brand color. Pass empty array to clear all tags.
orshot_update_brand_font_tags
ChatGPTReplace tags on a custom font. Pass empty array to clear all tags.
orshot_update_brand_image_tags
ChatGPTReplace tags on a brand asset image. Pass empty array to clear all tags.
orshot_update_brand_video_tags
ChatGPTReplace tags on a brand video. Pass empty array to clear all tags.
orshot_update_profile
ChatGPTUpdate the authenticated user's profile: full name, work role, and work/company URL. Use this during onboarding after asking the user about themselves. ONLY these three fields are writable; credit, trial and account-ownership fields (credits, trial counters, customer id, email) can never be changed through the API and return a permanent forbidden_fields refusal. Omit any field to leave it unchanged; pass an empty string to clear it.
orshot_update_social_post
ChatGPTChange a draft or scheduled post before it goes out — its caption, its media, which accounts it targets, or when it fires. The post keeps its ID. This is the review loop: draft a post, show the user, apply their edit here, then send it with orshot_publish_social_draft. Editing costs nothing and does not re-render anything. Only fields you pass are changed; everything else is kept. Passing status: "draft" turns a scheduled post back into a held draft. Works only while Orshot is still holding the post. Once it has gone to the platforms it is immutable — create a new post instead.
orshot_update_template_design
ChatGPTUpdate an existing studio template's design structure. Use this after orshot_create_template_design to add elements and build the design. You can also use this to update an existing template's design by fetching its current state with orshot_get_studio_template first, then sending modified pages_data. Skip that fetch for a template you created or edited in this conversation: that result's pageMap has every page_id, element_id and parameter_id, and its updatedAt is your base_updated_at. Edit specific pages instead of re-sending pages_data - page_operations: add, duplicate, replace, update settings (name, canvas, audioTracks, subtitle, videoDuration, motion), delete or move single pages. E.g. [{ "op": "duplicate", "page": 1 }], [{ "op": "update", "page_id": "…", "set": { "canvas": { "backgroundColor": "#000" } } }], [{ "op": "add", "page_data": { ...page } }]. Applied in order, all or nothing. - Name a page by page_id (stable, from pageMap) or page (1-based number). Prefer page_id: numbers shift when pages are added, deleted or moved. - Changing the number of an existing page (move, delete or insert before the last page) needs allow_reindex: true; renders that address pages by number then hit different pages. - Layer edits (add, remove or restack layers, SAVED animation timing and motion) → orshot_patch_template_elements layer_operations. - page_operations can't be combined with pages_data, name, description, tags or canvas size. Pass base_updated_at; the result's updatedAt is the value for your next write. Thumbnails are regenerated automatically in the background whenever pages_data, canvas_width, or canvas_height change. Pure metadata updates (name, description, tags) do not trigger re-rendering. You do NOT need to render the template manually to refresh the preview. Using Brand Assets Before updating, fetch relevant brand assets with orshot_get_brand_kit (filtered by tags matching your design purpose): - Brand colors → use their value field for style.color, style.fill, canvas.backgroundColor - Brand fonts → use the font name (without .ttf/.otf extension) as style.fontFamily. Custom workspace fonts take priority over Google Fonts. - Brand images → use their direct_url as element content for image elements - Brand videos → use their direct_url as element content for video elements 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(…
orshot_update_template_sharing
ChatGPTTurn public sharing of a studio template on or off, and control whether visitors can duplicate it. Returns the public link, embed page and iframe embed code. Fields you omit keep their current value. The link is minted once and stays stable — turning sharing off and on again revives the same link. allowDuplication is not available on every plan.
orshot_update_workflow
ChatGPTUpdate a workflow — rename, edit its steps, or change status (draft/active/paused/archived). Send only the fields you want to change. Activating requires the workflow to be complete and runnable.
orshot_update_workflow_sharing
ChatGPTTurn public sharing of a workflow on or off. Sharing publishes it to a public page anyone can view and copy into their own workspace. Fields you omit keep their current value. The link is minted once and stays stable — turning sharing off and on again revives the same link.
orshot_update_workspace
ChatGPTUpdate the active workspace: rename it (title), change its URL slug, and/or save onboarding answers. onboarding is a free-form JSON object that shallow-merges into what's already saved (set a key to null to delete it) — use it to record what the user wants to deliver, where, how often, and setup progress, so any later session (dashboard or agent) can resume. Slugs are lowercase letters/numbers/hyphens, 3-48 chars, globally unique; changing the slug changes all workspace URLs, so confirm with the user first.
orshot_upload_brand_audio
ChatGPTUpload an audio file to the brand library. Sources, exactly one per call: attachment = an audio file the user attached in this chat (the host fills it in where file inputs are supported; never invent a URL for an attached file); file = a public URL or base64. Supported: MP3, WAV, M4A, AAC, OGG, Opus, FLAC, WebM, AIFF, WMA. Max 50MB. For a local file (or a chat attachment your sandbox can read) use the direct flow instead of base64: orshot_create_upload_url (kind "audio") → PUT the bytes → orshot_confirm_upload. Use tags to categorize (e.g. ["music", "upbeat"], ["voiceover"]). Pass metadata.duration (seconds) when known — URL/base64 uploads cannot probe it automatically.
orshot_upload_brand_font
ChatGPTUpload a custom font to the workspace. Sources, exactly one per call: attachment = a font file the user attached in this chat (the host fills it in where file inputs are supported; never invent a URL for an attached file); file = a public URL or base64. Supported: TTF, OTF, WOFF, WOFF2. Max 10MB. For a local file (or a chat attachment your sandbox can read) use the direct flow instead of base64: orshot_create_upload_url (kind "font") → PUT the bytes → orshot_confirm_upload. Use tags to categorize (e.g. ["heading", "brand"], ["body", "regular"]).
orshot_upload_brand_image
ChatGPTUpload an image to brand assets and get a permanent direct_url for templates. Sources, exactly one per call: attachment = an image the user attached in this chat (the host fills it in where file inputs are supported; the way to ingest "my photo / my logo / this flyer"); file = a public URL or a base64 string; filePath = a local file (LOCAL stdio transport only). Never invent a URL for a file the user attached: pass it as attachment; if the host did not fill it in, use the direct upload flow (orshot_create_upload_url → PUT from your sandbox → orshot_confirm_upload) or ask for a public link. Do not re-type an attached file as base64: it overflows for anything but an icon. Supported: PNG, JPEG, WebP, SVG, GIF. Max ~7MB for base64/filePath (JSON body limit), auto-resized to 2000×2000 max (SVGs kept as-is). Use tags to categorize for easy retrieval when building templates (e.g. ["logo", "brand"], ["background", "dark"]).
orshot_upload_brand_video
ChatGPTUpload a video to brand assets. Sources, exactly one per call: attachment = a video the user attached in this chat (the host fills it in where file inputs are supported; never invent a URL for an attached file); file = a public URL or base64. Supported: MP4, WebM, MOV, AVI, MKV, FLV, WMV, M4V, GIF. Max 60MB. For a local file (or a chat attachment your sandbox can read) use the direct flow instead of base64: orshot_create_upload_url (kind "video") → PUT the bytes → orshot_confirm_upload. Use tags to categorize (e.g. ["intro", "brand"], ["background", "loop"]).
orshot_use_stock_media
ChatGPTResolve a stock ref from orshot_search_stock_media into a permanent Orshot-hosted URL and add it to the workspace Brand Library. Use it when you need the URL now (a render modification, orshot_place_image, a hand-off to the user). For template designs you can skip this and pass the ref as element content directly. Idempotent: the same ref resolves to the same file.
orshot_validate_workflow
ChatGPTDry-run a set of workflow steps WITHOUT saving — returns whether it's valid, runnable, plus any errors (blocking) and warnings (incomplete: missing config, no destination, no trigger). Use this to iterate on steps before orshot_create_workflow.
orshot_verify_design
ChatGPTRun the full design verification suite against a SAVED template's current state — the same geometry/shadow/encoding/font checks that run on save, on demand. Use it after a series of patches, before a final render, or whenever you want to confirm a design is clean without sending the design again. Checks: text occluded by higher-z elements, content off-canvas, dead/heavy shadows, un-encoded SVGs, misparsed "<" text, unavailable fonts (probed against the renderer's exact Google Fonts request), animation-timing and parameterization gotchas. Returns ✅ when clean, or the findings as text AND structuredContent.designWarnings (machine-readable, works on clients that can't see images).
orshot_add_brand_color
Claudeorshot_auth_status
Claudeorshot_batch_render
Claudeorshot_batch_render_template
Claudeorshot_configure_embed
Claudeorshot_confirm_image_upload
Claudeorshot_confirm_upload
Claudeorshot_create_dynamic_url
Claudeorshot_create_image_upload_url
Claudeorshot_create_template_design
Claudeorshot_create_template_folder
Claudeorshot_create_upload_url
Claudeorshot_create_workflow
Claudeorshot_delete_brand_audio
Claudeorshot_delete_brand_color
Claudeorshot_delete_brand_font
Claudeorshot_delete_brand_image
Claudeorshot_delete_brand_video
Claudeorshot_delete_dynamic_url
Claudeorshot_delete_template
Claudeorshot_delete_template_folder
Claudeorshot_delete_workflow
Claudeorshot_discard_social_post
Claudeorshot_duplicate_template
Claudeorshot_generate_image
Claudeorshot_generate_image_from_library
Claudeorshot_get_account
Claudeorshot_get_best_time_to_post
Claudeorshot_get_brand_audio
Claudeorshot_get_brand_colors
Claudeorshot_get_brand_fonts
Claudeorshot_get_brand_images
Claudeorshot_get_brand_kit
Claudeorshot_get_brand_videos
Claudeorshot_get_dynamic_url
Claudeorshot_get_embed_config
Claudeorshot_get_library_template_modifications
Claudeorshot_get_render_job
Claudeorshot_get_social_analytics
Claudeorshot_get_social_analytics_daily
Claudeorshot_get_social_followers
Claudeorshot_get_social_post
Claudeorshot_get_social_post_analytics
Claudeorshot_get_studio_template
Claudeorshot_get_studio_template_modifications
Claudeorshot_get_template_sharing
Claudeorshot_get_workflow
Claudeorshot_get_workflow_connection_data
Claudeorshot_get_workflow_run
Claudeorshot_get_workflow_sharing
Claudeorshot_get_workflow_webhook
Claudeorshot_get_workspace
Claudeorshot_import_brand_website
Claudeorshot_list_connected_integrations
Claudeorshot_list_library_templates
Claudeorshot_list_social_accounts
Claudeorshot_list_social_posts
Claudeorshot_list_studio_templates
Claudeorshot_list_template_folders
Claudeorshot_list_workflow_nodes
Claudeorshot_list_workflow_runs
Claudeorshot_list_workflows
Claudeorshot_list_workspace_logs
Claudeorshot_list_workspaces
Claudeorshot_move_template_to_folder
Claudeorshot_patch_template_elements
Claudeorshot_place_image
Claudeorshot_publish_social_draft
Claudeorshot_publish_social_post
Claudeorshot_rename_template_folder
Claudeorshot_render_library_template
Claudeorshot_render_template
Claudeorshot_reset_dynamic_url_signature
Claudeorshot_run_workflow
Claudeorshot_scan_brand_website
Claudeorshot_search_stock_media
Claudeorshot_set_active_workspace
Claudeorshot_undo_brand_import
Claudeorshot_update_brand_audio_tags
Claudeorshot_update_brand_color_tags
Claudeorshot_update_brand_font_tags
Claudeorshot_update_brand_image_tags
Claudeorshot_update_brand_video_tags
Claudeorshot_update_profile
Claudeorshot_update_social_post
Claudeorshot_update_template_design
Claudeorshot_update_template_sharing
Claudeorshot_update_workflow
Claudeorshot_update_workflow_sharing
Claudeorshot_update_workspace
Claudeorshot_upload_brand_audio
Claudeorshot_upload_brand_font
Claudeorshot_upload_brand_image
Claudeorshot_upload_brand_video
Claudeorshot_use_stock_media
Claudeorshot_validate_workflow
Claudeorshot_verify_design
Claude