add_to_cart
ChatGPTAdd a designed product to the user's VistaPrint shopping cart. The design is identified by its workId: the cart gets the work's latest saved design, including every edit_document change. The latest create_document, edit_document, or get_document response suggests complete add_to_cart args in next_suggested_actions. Returns the direct cart handoff in the top-level url field. When presenting the checkout link, use that exact url without modifying, truncating, re-encoding, or routing it through a search engine or third-party redirector. Subsequent edits via edit_document automatically update this cart item.
configure_product
ChatGPTLow-level building block: iteratively resolve a product configuration. With no selections, returns the full option set; with selections, resolves them against the current option set — the response reports the resulting completed selections and which options remain. Conflicting selections return an unresolvable state (not an error), with the remaining valid options for recovery. refine_product_selections combines this resolution with pricing in a single call and is the preferred tool for most configuration flows; this tool remains available for callers that want configuration state only.
create_document
ChatGPTCreate a design document for a product — the single source of panel IDs, dimensions, and sections for that product/options combination. The response's panels include each panel's id/name/width/height, text/image placeholders and purposes, and semantic section data, useful for sizing and targeting artwork in external design tools before editing. The top-level url is the exact Studio handoff link; present it without modifying, truncating, re-encoding, or routing it through a search engine or third-party redirector. previews holds one image per printable side of this design; preview URLs are opaque, so never construct or edit one; after edits, call get_document or render_preview for current ones. Two modes: (A) provide a templateToken from search_templates to start from a template, or (B) omit templateToken to get a blank full-bleed document with placeholder images. edit_document customises text, style, and images using operations against these panel IDs; render_preview and add_to_cart operate on the resulting workId.
edit_document
ChatGPTEdit a design document using a sequence of operations. Supports: editText (change text content), editTextStyle (change font, size, color, alignment), replaceImage (swap image URL), addText (add new text area), addImage (add new image), move (reposition/resize), reorder (change z-index), remove (delete element). Use the panel summary from create_document or get_document to see current field IDs. Operations execute in order. If this design is in the cart, the correlated cart item is updated automatically; add_to_cart uses the latest saved design for this workId. The response previews are per-side images of the edited design; present them unchanged and never construct or edit one. The top-level url is the Studio link. render_preview reflects the result on the product.
get_document
ChatGPTRetrieve a summary of a Work entity's design — text fields, images, shapes (with style, rotation, and z-index), and previews (one ready-made image per printable side, labelled e.g. Front/Back, for the current document revision). Preview URLs are opaque: present them exactly as returned and never construct or edit one; call get_document or render_preview for current ones. Does NOT return raw CimDoc.
get_price
ChatGPTLow-level building block: price one or more option×quantity combinations for a product in a single call — tax-aware, with product/shipping/item total breakdowns. Advisory/display-only: does not create or feed a cart. refine_product_selections combines configuration resolution with this pricing logic in a single call and is the preferred tool for most configuration flows; this tool remains available for pricing an already-fullyResolved configuration across a matrix of options and/or quantities directly.
list_projects
ChatGPTList the user's saved designs from their VistaPrint account (My Projects), newest first — the source for reprinting, editing, or reordering a past design. Returns up to 9 projects; narrow with name (matches part of the project name) or productKey when the user has described the design they mean. When hasMore is true, repeat the call with the same filters and offset set to nextOffset for the next page. Each project has a workId usable with get_document or edit_document. Each project's url is the exact Studio handoff link; present it without modifying, truncating, re-encoding, or routing it through a search engine or third-party redirector.
product_details
ChatGPTDiscover available options and quantities for a product, plus a sensible, quantity-compatible default configuration, given a productKey/productVersion from search_products. refine_product_selections resolves the final selection from these options and prices it in the same call.
refine_product_selections
ChatGPTPreferred high-level tool for narrowing a product configuration and pricing it in the same call. Given partial/updated option selections, resolves them (like configure_product) and reports the resulting state — fullyResolved, partiallyResolved, or unresolvable — plus completed selections and remaining valid options. Prices requested pivots (default: Quantity) when the configuration is resolved enough; when it is not, returns configuration state plus an unpricedReason instead of guessing at a price. configure_product and get_price remain available as lower-level building blocks.
remove_asset_background
ChatGPTRemove the background from an existing Sherbert asset (assetId from upload_asset), returning a linked, ready-to-use transparent variant. Processing can be asynchronous — a processing status is not an error; retry this same tool after retryAfterMs.
render_preview
ChatGPTRender an interactive product preview (3D where available, plus 2D scenes and flat per-side views) of a design on its product, using the workId from create_document. The interactive preview shows the user every available view. The response has previews, one flat image per printable side labelled by name (e.g. Front, Back), and a top-level url for the 2D product scene, all for the current document revision. Preview URLs are opaque: present them exactly as returned and never build or edit one. A side that is not in previews has no preview; say so rather than constructing a URL. After edits, call render_preview again for current URLs.
request_product_selection
ChatGPTDisplay a focused shortlist of catalog-resolved products and wait for the user to type their choice. This tool is the required user-input checkpoint when multiple viable candidates remain, the user is ready to proceed, and their preference remains unresolved, including requests to see, compare, or choose among candidates. A prose recommendation does not replace this checkpoint. Input is a focused shortlist, preferably 3 to 6 products, naming only each candidate's productKey, productVersion, and a reason it fits; this tool resolves display fields and the View link from the catalog itself, not from caller-supplied data. Candidates the catalog cannot resolve are omitted and reported with their requested position; the remaining products are renumbered consecutively. Background research, requests to postpone choosing, empty results, one exact match, and an already-selected product are outside its scope. This tool is terminal for its response; it names product_details as the next step, with no arguments, since that call resumes only once the user types their choice.
search_assets
ChatGPTList the user's own previously uploaded or accessible image assets, newest first by default. Returns print-ready URLs usable in edit_document operations (replaceImage or addImage). There is no keyword search — filter by date range or recency only.
search_products
ChatGPTSearch VistaPrint products by keyword for catalog discovery and downstream tool inputs. This tool returns data without selection UI. Each product url is an absolute product-page link for direct handoff. For multiple viable results when the user is ready to proceed and their preference remains unresolved, request_product_selection is the correct next interaction, naming only each candidate's productKey, productVersion, and a reason it fits; that tool resolves display fields and its own View link from the catalog, so platformUrl need not be carried forward. A prose recommendation does not resolve the required user-input checkpoint. Background research and requests to postpone choosing remain data-only, even with multiple results. Empty results and one exact clear product also bypass selection UI. Returns name, productKey, productVersion, priceFrom, imageUrl, options, galleryName, mpvId, url, and platformUrl.
search_templates
ChatGPTSearch VistaPrint's template gallery for pre-designed templates. Only applicable when the product from search_products has a galleryName; products without one have no templates, so artwork comes from an external design tool via upload_asset instead. selectedOptions from configure_product filters templates to the user's confirmed choices (e.g., substrate, orientation, shape). Keywords narrow the gallery to the user's style, industry, or occasion; without them results are generic defaults. Returns up to 9 templates with preview images and templateTokens for create_document. When hasMore is true, repeat the same search with offset set to nextOffset to show the next 9.
upload_asset
ChatGPTImport artwork — from a public HTTPS URL, a file attached in the current chat, or (in Gemini) server_img_id, a server-side identifier for an image generated earlier in the conversation — and return a print-ready asset URL for use in edit_document operations (replaceImage or addImage). In Gemini, when the user asks to print or use an image generated earlier in the chat, pass its identifier in server_img_id instead of asking for an external URL.
