Canva

Gets the dataset schema of a brand template. The schema contains available autofill fields and their data types (text, image, etc.). The schema is used to understand what dynamic content can be autofilled into the template. If this tool returns an empty object, it means the template does not support autofill, and you could suggest the user to create a Canva design using the brand template manually by clicking the `create_url` field of the brand template. The `create_url` can be found by the `search-brand-templates` tool.

get-brand-template-dataset

Internal tool for widget UI polling.

get-design-candidates

Get detailed information about a Canva design, such as a doc, presentation, whiteboard, video, or sheet. This includes design owner information, title, URLs for editing and viewing, thumbnail, created/updated time, and page count. This tool doesn't work on folders or images. You must provide the design ID, which you can find by using the `search-designs` or `list-folder-items` tools. You may also pass a full Canva design share URL directly (do not pre-extract just the ID) — any collaboration token embedded in the URL is needed to access a design shared via link and will be parsed and forwarded automatically. Example URL: https://www.canva.com/design/{design_id}. If the user provides a shortlink (e.g. https://canva.link/abc123), use `resolve-shortlink` with the shortlink ID first to get the full URL.

get-design

Get the thumbnail for a particular page of the design in the specified editing transaction. This tool needs to be used with the `start-editing-transaction` tool to obtain an editing transaction ID. You need to provide the transaction ID and a page index to get the thumbnail of that particular page. Each call can only get the thumbnail for one page. Retrieving the thumbnails for multiple pages will require multiple calls of this tool.IMPORTANT: ALWAYS ALWAYS ALWAYS show the preview to the user of EACH thumbnail you get in the response in the chat, EVERY SINGLE TIME you call this tool

get-design-thumbnail

Get a list of pages in a Canva design, such as a presentation. Each page includes its index and thumbnail. This tool doesn't work on designs that don't have pages (e.g. Canva docs). You must provide the design ID, which you can find using tools like `search-designs` or `list-folder-items`. You may also pass a full Canva design share URL directly (do not pre-extract just the ID) — any collaboration token embedded in the URL is needed to access a design shared via link and will be parsed and forwarded automatically. You can use 'offset' and 'limit' to paginate through the pages. Use `get-design` to find out the total number of pages, if needed.

get-design-pages

Get the text content of a doc, presentation, whiteboard, social media post, and other designs in Canva (except sheets, as it does not return data in sheets). Use this when you only need to read text content without making changes. IMPORTANT: If the user wants to edit, update, change, translate, or fix content, use `start-editing-transaction` instead as it shows content AND enables editing. You must provide the design ID, which you can find with the `search-designs` tool. You may also pass a full Canva design share URL directly (do not pre-extract just the ID) — any collaboration token embedded in the URL is needed to access a design shared via link and will be parsed and forwarded automatically. Do not use web search to get the content of a design as the content is not accessible to the public. Example URL: https://www.canva.com/design/{design_id}.

get-design-content

Internal tool for widget UI polling.

get-design-tool-job

Get the presenter notes from a presentation design in Canva. Use this when you need to read the speaker notes attached to presentation slides. You must provide the design ID, which you can find with the `search-designs` tool. When given a URL to a Canva design, you can extract the design ID from the URL. Example URL: https://www.canva.com/design/{design_id}.

get-presenter-notes

Allows a user to share a flat image file (PNG, JPEG, or WEBP) and convert it into a fully editable Canva design. Images can be provided via a public URL, a server-side image ID, a platform file reference, or via the Magic Layers upload tile. Best suited for: - Posters - Banners - Flyers - Social graphics - Marketing collateral - Other flat, non-photo-realistic designs created outside Canva **When to use:** Call this tool when the user: - asks whether it's possible to edit elements inside an image This applies: - even if they do not upload or share a link to the image → make sure to also include a note with what it is best suited for - if they upload or provide a public URL to a: - non-photo realistic PNG, JPEG, or WEBP file (e.g. illustrations, graphics, posters, banners) - photo-realistic PNG, JPEG, or WEBP file (e.g. people, animals, landscapes) → make sure to include a strong caveat that this feature works best for flat designs - if the user has shared an image and a server_img_id was provided → use the server_img_id parameter **IMPORTANT – Re-invocation rule:** Each call to this tool starts a completely fresh, independent workflow: a new widget instance is created, and a new design_id will be produced. A widget rendered by a previous tool call in this conversation is no longer active. Do NOT skip calling this tool on the assumption that an earlier widget is still usable. **When NOT to use:** Don't call this tool when the user: - provides a public URL to an image, but does not explicitly ask to edit it → use upload-asset-from-url - uploads images as inspiration only (does not explicitly ask to edit them) → use generate-design - asks to edit an existing Canva design → use perform-editing-operations **When to provide the url parameter:** Provide the url parameter ONLY when the user explicitly provides a public URL string for an image. DO NOT provide the url parameter if you only have a local or container path. **When to provide the server_img_id parameter:** Provide the server_img_id parameter when a server-side image identifier has been provided. Use server_img_id instead of url in this case; do not provide both. **When to provide the image_file parameter:** Provide the image_file parameter when a platform file reference object has been supplied (e.g. a ChatGPT-uploaded or GPT-generated image). Use image_file instead of url in this case; do not provide both. **Workflow:** If the user provides a public URL: 1. Start processing the image immediately using the url parameter. If a server_img_id has been provided: 1. Start processing the image immediately using the server_img_id parameter. If a platform file reference has been provided: 1. Start processing the image immediately using the image_file parameter. If the user does not provide a public URL, a server_img_id, or a platform file reference: 1. Open the Magic Layers upload widget: ui://widget/image-to-design.html 2. The user must upload the image by: - using the tile's "Select media" button, or - dragging-and-dropping it onto the tile 3. If the user uploaded an image before this tool was called, they must re-upload it through the tile 4. After the file has been converted to an editable design, a new design_id is created and can be used with: - start-editing-transaction, then perform-editing-operations - resize-design - export-design **Behaviour:** - A new Canva design and design_id is created on every tool call - The flat file is separated into editable elements, which are all customizable - The resulting design can be opened in Canva for further editing **Examples:** Use this tool when the user says: - "I want to be able to edit the text in this image (no upload or description of the image provided)" - "I want to change the color of my dog's collar in this photo" - "Call the image to design tool" or "Run image-to-design again" → always invoke the tool; do not reuse a widget or result from earlier in the conversation Do not use when the user says: - "Here's a link to an image I want to save to Canva." → use upload-asset-from-url - "I like this kind of poster and want to create something similar" → use generate-design

image-to-design

ALWAYS use this tool when the user's message contains an HTTPS URL, or a file generated/uploaded in this chat, and their intent is to create a Canva design from it. Pass public HTTPS URLs directly via url. Use design_file for chat-generated or uploaded files, including HTML files and ZIP bundles. **When to provide the design_file parameter:** Provide design_file when a platform file reference object has been supplied, or when the user wants to import a file artifact generated/uploaded in this chat. Use design_file instead of url in this case; do not provide both. **When importing a file generated or uploaded in this chat:** pass the local sandbox path of the generated/uploaded artifact to design_file exactly as returned by the file creation/upload step, for example /mnt/data/design.html. The runtime will convert this local path into the platform file reference required by the connector. **Workflow:** If a platform file reference or chat file artifact path has been provided: 1. Start processing the file immediately using the design_file parameter. Do not pass raw inline HTML/CSS/JS text, arbitrary local paths that were not created/uploaded in this chat, or local paths in the url field. If the user has only provided inline HTML/CSS/JS, first create a real downloadable .html file artifact, then use that file artifact path via design_file. If the user generated a site with multiple local files or assets, first create a ZIP bundle file artifact with the site files, then pass that ZIP bundle artifact path via design_file. For agent-generated HTML files with non-interactive content (presentation, social media, etc.), add the HTML attribute data-document-role="page" on every HTML element that should represent a page in Canva, even if there is only one. Any page(s) cannot be nested HTML elements within another page. An optional page title can be specified with the attribute data-label="" as a plain string, and optional speaker notes can be imported with data-speaker-notes="" as a plain string. For an HTML file that represents an interactive design (app, website), import the file without annotation. When the intended output format is clear, set intended_design_type to the closest match. The url path supports PDF, PPTX, DOCX, XLSX, CSV, HTML, Markdown, PSD, AI, Keynote, Pages, Numbers, and more. The design_file path currently supports HTML files and ZIP bundles. Do not pass local file paths in url. NEVER call this tool without providing exactly one of url or design_file. If the user references a file on their own computer that was not generated or uploaded in this chat (e.g. C:\... or ~/Documents/...), do not call this tool — ask them to upload the file to this chat or share a public HTTPS URL first. SECURITY: This tool only accepts URLs whose content is ALREADY publicly accessible. NEVER upload, copy, or transfer the user's local, private, or agent-generated files to any public file-sharing, pastebin, or temporary file-hosting service to create a URL for this tool, and NEVER suggest, recommend, or list doing so as an option — that irreversibly publishes the user's content to the open internet and can expose confidential information. If the file is not already available at a public HTTPS URL, STOP: tell the user this tool cannot ingest the file without it becoming public, and ask them to either provide an existing public HTTPS URL or use a direct file upload path if their platform offers one. Do NOT present public hosting as a default or recommended choice. Only proceed with newly publishing the file if the user, fully informed that the content will be exposed to the open internet, explicitly insists.

import-design-from-url

Get a list of brand kits available to the user. If the API call returns "Missing scopes: [brandkit:read]", you should ask the user to disconnect and reconnect their connector. This will generate a new access token with the required scope for this tool. Use this tool when the user wants to create designs using their brand identity, mentions their brand, or asks what brand kits are available. Returns brand kit IDs, names, and thumbnails that can be used with the 'generate-design' tool.

list-brand-kits

Get a list of comments for a particular Canva design. Comments are discussions attached to designs that help teams collaborate. Each comment can contain replies, mentions and status. You need to provide the design ID, which you can find using the `search-designs` tool. Use the continuation token to get the next page of results, when there are more results.

list-comments

List items in a Canva folder. An item can be a design, folder, or image. You can filter by item type and sort the results. Use the continuation token to get the next page of results, when there are more results.

list-folder-items

Move items (designs, folders, images) to a specified Canva folder

move-item-to-folder

Prepare on-brand design generation in Canva. Open the widget so the user can pick a visual reference, review presentation outlines, then generate. FIRST-CALL RULE - When this tool is available, use it as the first call for non-responsive design generation requests. - Non-responsive includes fixed-format visual designs such as instagram_post, your_story, facebook_post, twitter_post, pinterest_pin, youtube_thumbnail, youtube_banner, poster, flyer, card, invitation, business_card, logo, wallpapers, photo_collage, infographic, and postcard. - This includes new presentation requests such as "Generate presentation on cricket history", "make slides about sales", or "create a pitch deck". - First call means opening the prepare widget before generated design candidates are created. - For the first call, omit intent or set it to recommendation. - Include design_type and topic in the first call. - If design_type is presentation, include outline in the same call. WORKFLOW - Use intent recommendation (default) to fetch or hydrate the widget's visual-reference recommendation. - If the conversation mentions a brand kit, pass its ID as brand_kit_id. - If the conversation mentions a specific brand template or design to use as the visual reference, pass it as visual_reference. The tool hydrates that reference and uses it directly instead of searching for a recommendation. - Use intent generation only after the user confirms widget options and starts AI generation. - Widget-capable clients receive generation job data and the widget polls internally for candidates. RESPONSE PLACEMENT - For intent recommendation, put any user-facing setup text before calling this tool. - After a recommendation call, do NOT add assistant text below the widget. The widget is the final visible response for that turn. REQUIRED INPUT - topic: what to create (subject, goals, brief). Must be a non-empty string; if the user gives no explicit topic, derive a concise brief from their request. Never pass an empty string. - outline: required whenever design_type is presentation, including the first recommendation call REQUIRED FOR GENERATION INTENT - design_type: e.g. presentation, instagram_post, flyer OPTIONAL INPUT - title: a concise title for the generated Canva design; when omitted, it is derived from topic - brand_kit_id, visual_reference, source_document, audience, length - Prefer visual_reference for model-facing calls. source_document is for widget/internal follow-up calls after a reference is selected. OUTLINE REVIEW - Presentations include fullscreen outline review in this widget. - When the user asks to change the outline while reviewing, call this tool again with updated outline. - Do NOT call request-outline-review during this flow; that tool is for the legacy guided presentation flow only. SINGLE-PAGE DESIGN TYPES - Presentations are multi-slide: describe all slides via the outline. - Every other design type produces a SINGLE page. Describe one page only. Do NOT describe multiple pages, sides, or faces (e.g. the front and back of a business card); multi-page descriptions produce poor results. WHEN NOT TO USE - Responsive or document-like outputs: docs, documents, sheets, spreadsheets, videos, video-first designs, email designs, full websites, or import-from-URL flows. - Use topic and outline for the content brief. CHAT IMAGE VISUAL REFERENCE - When the user refers to an image uploaded or generated in this conversation, pass its platform file reference as **image_file**, including for presentations. - For non-presentation requests, **image_file** is the visual reference and starts generation in the widget immediately. - For presentations, still pass **image_file** when the user referred to a chat image. The widget explains that presentations cannot use it as a visual reference. - Continue to use **visual_reference** when the user refers to an existing Canva design or brand template. Do not use the widget/internal **reference** or **source_document** inputs for model-facing recommendation calls.

prepare-design-generation

Resize a Canva design to a preset or custom size. The tool will provide a summary of the new resized design, including its metadata.

resize-design

Resolves a Canva shortlink ID to its target URL. IMPORTANT: Use this tool FIRST when a user provides a shortlink (e.g. https://canva.link/abc123). Shortlinks need to be resolved before you can use other tools. After resolving, extract the design ID from the target URL and use it with tools like get-design, start-editing-transaction, or get-design-content.

resolve-shortlink

Commit an editing transaction. This will save all the changes made to the design in the specified editing transaction. CRITICAL: All edits are in DRAFT and will be PERMANENTLY LOST if this tool is not called. You MUST always show the user what changes were made and ask for their explicit approval before calling this tool — for example: "Would you like me to save these changes to your design?" Wait for their clear approval before proceeding. Do NOT call this tool without user approval. After successfully saving changes, always provide the user with a direct link to open their design in Canva for review. Use the link they gave you or from the get-design tool. If the commit fails, ALL changes made during the transaction are lost and no changes are saved to the actual design. Users must start a new editing transaction to retry any failed operations. Once an editing transaction has been committed, the `transaction_id` for that editing transaction becomes invalid and should no longer be used.

commit-editing-transaction

Searches and lists for templates available to the current user. This returns a list of templates with their IDs, names, and preview thumbnails to help users select the appropriate template for their needs. ** ❗When a user says search a template, they ALWAYS mean brand-templates. Therefore ALWAYS call this tool to search for the templates ** ** 🟢 This tool can also be used to browse all of the available templates. If query is not provided, all available templates will be returned. ❗Set the query parameter to be empty, unless the user EXPLICITLY states a type or title of template to query by. You may use dataset="non_empty" with or without a query. When used without a query, this returns all autofill-capable templates ** Use the continuation token to get the next page of results, if needed. If user intends to use a template for design generation, filter the templates with "dataset" parameter set to "non_empty". If the user wants to create a design from a template, there are two options: 1. Design Generation from template. Only templates with non-empty dataset can be used for design generation. You also need to set the dataset parameter to non-empty to get the templates for design generation. Then use the 'autofill-design' tool to generate. 2. Create a design manually in the browser. If you cannot find a suitable template for design generation, you can suggest the user create a Canva design using the template manually, by going to the 'create_url' field of the template in the browser. ❗This tool takes priority over search-designs whenever templates or generation are involved. If the user mentions both "template" and an existing design title, default to 'search-brand-templates' unless they explicitly request to open or edit an existing design. ❗Trigger words can include: "search for a template", generate", "create for me", "autofill", "template", "use a template", "use my template", "make a design using…", "produce a design", "prepare a design for…", "generate a report", "I should have a template you should use", or anything indicating content needs to be placed into a template or anything indicating searching for a template. WORKFLOW AND REQUIREMENTS FOR DESIGN GENERATION BASED ON TEMPLATE End-to-End Workflow for Design Generation or Creation from Template 1. User asks that they want to "generate or create a design from a template". 2. System calls this tool with non-empty dataset parameter. 3. System returns a list of fillable templates to generate or create designs from. 4. User selects one of the templates. 5. System calls 'get-brand-template-dataset' tool to confirm that the designs can be created from the template (ie. its non-empty) 6. System calls this tool with that template, once the template is confirmed by the user. 7. System returns the new design created from the template. Critical Requirements: 1. ALWAYS use this tool whenever the user wants to generate any new design from a template. 2. ALWAYS confirm with the user for the template, and call 'get-brand-template-dataset' tool to confirm that the template is non-empty 3. ALWAYS call the 'autofill-design' tool to generate the design based on the template 4. ** 🟢 ALWAYS use this tool when the user expresses intent to "generate", "create", "autofill", "search a template", "start from a template", "use my template", or "pick a template for generation". In all such cases, ALWAYS use search-brand-templates. ANY query involving: – "generate a presentation" – "generate a report" – "make a design using a template" – "generate from a template" – "produce a presentation from their template" - "search for available templates" MUST ALWAYS use search-brand-templates.**

search-brand-templates

Search docs, presentations, videos, whiteboards, sheets, and other designs in Canva. Use the continuation token to get the next page of results, if needed. The design URLs are secured and are not accessible to the public. Use the fetch tool instead of web search to get the content of a design. Use the continuation token to get the next page of results, when there are more results.

search

Search docs, presentations, videos, whiteboards, sheets, and other designs in Canva, except for templates or brand templates. Use when you need to find specific designs by keywords rather than browsing folders. Use 'query' parameter to search by title or content. If 'query' is used, 'sortBy' must be set to 'relevance'. Filter by 'any' ownership unless specified. Sort by relevance unless specified. Use the continuation token to get the next page of results, when there are more results. CRITICAL REQUIREMENTS: 1. ALWAYS use the 'search-brand-templates' tool when the user is searching for templates or wants to use a template. 2.** 🚫 When a user says search a template, they ALWAYS mean brand-templates. Therefore NEVER call this tool, ALWAYS call the 'search-brand-templates' tool to search for the templates. ** 3.** 🚫 NEVER use this tool when the user expresses intent to “generate”, “create”, “autofill”, “search a template”, “start from a template”, “use my template”, or “pick a template for generation”. In all such cases, ALWAYS use search-brand-templates. ANY query involving: – “generate a presentation” – “generate a report” – “make a design using a template” – “generate from a template” – “produce a presentation from their template” - "search for available templates" MUST NOT use search-designs. This tool ONLY searches existing designs (docs, presentations, whiteboards, videos, etc.) that the user already owns or that are shared with them. It DOES NOT find templates and MUST NOT be used as a fallback for template selection. **

search-designs

Search the user's folders and folders shared with the user based on folder names and tags. Returns a list of matching folders with pagination support. Use the continuation token to get the next page of results, when there are more results.

search-folders

Start an editing session for a Canva design. Use this tool FIRST whenever a user wants to make ANY changes or examine ALL content of a design, including:- Translate text to another language - Edit or replace content - Update titles - Connect or remove autofill field labels on existing text or image elements - Replace or insert media (images/videos) - Delete media/text - Fix typos or formatting - Format text appearance (color, alignment, decoration, links, lists, font (size, weight, style; family not supported)) - Auditing or reviewing content - Edit words or sections of a paragraph or text element.- Edit specific words or substrings throughout the Canva design CRITICAL REQUIREMENTS: 1. The `transaction_id` returned in the tool response MUST be remembered and MUST be used in all subsequent tool calls related to this specific editing transaction. 2. After calling this tool, ALWAYS call `perform-editing-operations` with the edit operations needed to apply the edits the user wants. To re-iterate, editing operations must be performed by the `perform-editing-operations` tool. 3. IMPORTANT: ALWAYS call the `perform-editing-operations` tool with the edit operations needed to apply the edits the user wants. 4. IMPORTANT: After performing ALL operations, ALWAYS call `commit-editing-transaction` to save the changes. Changes are in DRAFT only until committed — they will be PERMANENTLY LOST if you do not call `commit-editing-transaction`. 5. To save the changes made in the transaction, use the `commit-editing-transaction` tool. 6. To discard the changes made in the transaction, use the `cancel-editing-transaction` tool`.7. IMPORTANT: ALWAYS ALWAYS ALWAYS show the preview to the user of EACH thumbnail you get in the response in the chat, EVERY SINGLE TIME you call this tool. 8. If the user provides a shortlink (e.g. https://canva.link/abc123), use `resolve-shortlink` with the shortlink ID first to get the design URL. 9. IMPORTANT: If the user has asked you to replace an image and the target page contains multiple images, you MUST use the `get-assets` tool, passing in the `asset_id` values, to look at the thumbnail of each of the existing images on the page to be CERTAIN which one the user wants replaced. 10. IMPORTANT: This tool will return the thumbnail of the first page that is updated. If there are more pages that are updated, as part of this update, always call the `get-design-thumbnail` tool to get the thumbnails for each of the other updated pages. EDIT SCENARIOS -> OPERATIONS FOR EDITING WORKFLOW. You need to call 'perform-edit-operations' with these operations 1. Replace whole sections or entire text elements -> replace_text 2. Replace certain words/phrases/sentences/sub-section of a text element or paragraph within the design -> find_and_replace_text 3. Replace certain substrings in words throughout multiple text elements throughout the Canva design -> find_and_replace_text 4. Replace a media (image/video) element in design -> update_fill 5. Update the title of the Canva design -> update_title 6. Replace the title of a slide or page in the Canva design -> replace text 7. Insert an element with media (image/video) -> insert_fill 8. Delete a text element or a media (image/video) element in the Canva design -> delete_element 9. Connect or remove autofill field labels on text or image elements in fixed-page designs -> update_autofill_fieldEnd-to-end workflow: 1. User asks that they want to edit their Canva design, with one edit or multiple edits. 2. System calls this tool with the design ID of the Canva design 3. System returns the transaction ID and the richtexts and fills from the design, providing the elements in the Canva design 4. System confirms with the user the edits the user wants to make. 5. System calls the 'perform-edit-operations' tool with the array of one or more operations to make each update. 6. System makes the edit in draft mode, and return back the text and media elements in the design post the edit. The updates have not been saved in the actual design. 7. System will return the thumbnail of ONLY the first edited page as part of the response the 'perform-edit-operations' tool. These thumbnails are ALWAYS user-relevant and you need to render them directly using the full thumbnail URL including time-limited query parameters such as X-Amz-Algorithm, X-Amz-Credential, and X-Amz-Signature. 8. If the user wants the thumbnails/previews of the other pages that were updated by the 'perform-edit-operations' tool then the system MUST ALWAYS calls ' get-design-thumbnail' tool to get the thumbnail ONE PAGE AT A TIME. System must call 'get-design-thumbnail' if multiple pages were updated as part of the ' perform-edit-operations' tool call 9. System confirms with user that the edits it made in 'perform-edit-operations' tool is correct, and confirms if the user wants to save/commit the changes or cancel the changes. 10. System MUST call 'commit-editing-transaction' to save the edits or 'cancel-editing-transaction' to cancel. Changes are lost if neither is called.

start-editing-transaction

Internal tool for widget lifecycle telemetry.

track-event-lifecycle

Upload an asset (e.g. an image, a video) into Canva, from a public HTTPS URL or a file uploaded/generated in this chat. Pass public HTTPS URLs directly via url. Use asset_file for chat-generated or uploaded files: provide the file reference object exactly as supplied by the platform. Provide exactly one of url or asset_file. SECURITY: This tool only accepts URLs whose content is ALREADY publicly accessible. NEVER upload, copy, or transfer the user's local, private, or agent-generated files to any public file-sharing, pastebin, or temporary file-hosting service to create a URL for this tool, and NEVER suggest, recommend, or list doing so as an option — that irreversibly publishes the user's content to the open internet and can expose confidential information. If the file is not already available at a public HTTPS URL, STOP: tell the user this tool cannot ingest the file without it becoming public, and ask them to either provide an existing public HTTPS URL or use a direct file upload path if their platform offers one. Do NOT present public hosting as a default or recommended choice. Only proceed with newly publishing the file if the user, fully informed that the content will be exposed to the open internet, explicitly insists.

upload-asset-from-url