MCP App Store
by KeenEthics
Crayonz icon

Crayonz MCP Server for ChatGPT

By AIVOLVIX PRIVATE LIMITED

Overview

Crayonz turns an idea in a conversation into a printed garment. Describe a design, or attach your own artwork or photo, and Crayonz creates print-ready artwork and shows it on the real product, front and back, in the colours that garment comes in. What you can do: - Choose between two design options and refine them in plain words ("make the moon red", "add Team Rocket 2026 in gold"). - Add text, logos and shapes as separate layers, move the print to the back, a pocket or a sleeve, and preview the design across colours. - Browse the catalogue: classic and oversized tees, acid-wash tees and hoodies, polos, sweatshirts, zip-ups, a varsity jacket, caps, mugs, bottles and posters. - Check the size chart and choose size and quantity. When you're ready, Crayonz creates a checkout link. You pay on crayonz.ai in your browser; nothing is charged inside the chat. Orders are printed in India and ship across India and to 25 other countries. Crayonz signs in with your Crayonz account. It reads and writes only your own drafts, saved styles, logos and orders. Sexual, hateful or exploitative designs are refused.

How to connect

  1. ChatGPT: Open the Crayonz listing in the ChatGPT app directory and choose Connect. ChatGPT listing Crayonz exposes 50 tools here.

Tools

add_shape

ChatGPT
Adds a vector shape (circle, star, heart, triangle, rectangle; badge, burst, frame, banner, arrow and diamond approximate these) to an existing draft as a layer. Rendered as SVG in about 1-2 seconds; involves no AI image generation. Position is relative to the existing design: around, behind, top, bottom, left, right or center. Set fill for a solid shape, or stroke and strokeWidth for an outline (no fill gives an outline only). Freeform shapes such as paint splatter are drawn by modify_design, which draws with AI.

add_text

ChatGPT
Adds text to an existing draft as a layer, without regenerating the artwork. Renders exact letterforms as SVG on top of the current design and recomposes the mockup in about 1-2 seconds; involves no AI image generation. Styles: bold, outlined, script, vintage, modern, graffiti, minimal, retro. Color is a hex value (#RRGGBB) or CSS name; the default is white on dark garments and near-black on light ones. strokeColor and strokeWidth add an outline. Position: overlay (default, lower third of the design), above, below, left, right, or replace (text only, no underlying design). Curved text and multiple text blocks per design are available in the Studio via view_in_studio_url.

analyze_design

ChatGPT
Returns the palette, style, mood, composition and extracted text of an existing draft's current design. Read-only: it does not modify the draft and involves no AI image generation. Useful for describing what a draft looks like, or for checking its colours before an edit. For improvement ideas see get_design_suggestions. For an image URL that is not a draft, see analyze_reference_image.

analyze_reference_image

ChatGPT
Analyses an image from a URL (not a draft) with a vision model and returns its palette, style, mood, composition and extracted text. Read-only. Useful when a reference image is attached: the URL can then be passed to generate_design as reference_image_urls, or to blend_references when there are several. For an existing draft, use analyze_design.

apply_style

ChatGPT
Adjusts a draft's design toward a saved brand kit: builds a modify_design instruction from the preset's palette, font style and notes. Draws a new image with AI. view defaults to the front; 'back' applies to the back print of a multi-view draft. The change is not free to undo: reset_layers removes overlays, and list_versions with rollback_to_version restores an earlier version.

apply_suggestion

ChatGPT
Applies one suggestion from get_design_suggestions by passing it to modify_design. Input: suggestion_text, the suggestion as returned or a paraphrase. Observation-style text such as 'low contrast' is turned into an instruction such as 'Improve contrast: low contrast'. Draws with AI, the same as modify_design. view defaults to the front; 'back' targets the back print of a multi-view draft.

blend_references

ChatGPT
Generates a new draft that combines 2-4 reference images and returns it. The prompt states what to take from each image, for example 'palette from #1, shapes from #2'. Garment, colour and placement parameters match generate_design. For one reference image, generate_design accepts reference_image_urls; for an exact copy of one image it accepts seed_image_url; for a marketplace item see remix_listing.

check_job

ChatGPT
Returns the result of a background generation job, given its job_id. Generation that has started in the background is already charged once; this call collects the result without generating again. If the job is still queued or running, the response says so and the same id can be checked again a few seconds later.

clone_design

ChatGPT
Copies a draft's artwork onto a different garment or colour and creates a new draft with its own draft_id and a recomposed mockup. The source draft is unchanged. No AI image generation. A back print on the source is copied too; it is dropped when the target garment has no back view, and the response says so. If the target is zipper_hoodie (no front-center placement), the front design goes to left-pocket unless placement is given.

color_explorer

ChatGPT
Composes an existing draft in 2-8 colours in one call and returns permanent URLs for the front and/or back mockup in each colour. Colours can be slugs, display names (any case) or hex values, with nearest-colour matching and spelling suggestions. Takes about 2-3 seconds for 5 colours x 2 views. Repeating the call on an unchanged draft returns the same URLs. No AI image generation.

compose_layered_design

ChatGPT
Builds a draft from layout slots in one call: a front pocket badge, back-top text, an optional back-middle logo and optional back-bottom text. Every slot is optional. Deterministic and involves no AI image generation unless ai_base_prompt is set. Returns front and back mockups in about 4 seconds. Back slots: top (upper 40%), middle (middle 30%), bottom (lower 40%). The pocket badge defaults to scale 0.9; pocket_side 'right' uses the right chest. A logo is a search_logos slug or an allowlisted public image URL; text presets come from list_text_presets. Later edits: add_text, place_logo, update_layer.

create_checkout

ChatGPT
Holds the order for one draft and returns a single-use secure checkout link, valid for 30 minutes. The buyer opens the link to enter an address and pay: India in rupees (UPI, cards, wallets), other countries in their own currency with PayPal, shipping shown there. Inputs: draft_id; size (required for garments with more than one size, in the buyer's own words); quantity 1-5. Caps, mugs, bottles and posters are single-size. The print-resolution file is prepared before the link is created. Delivery is 4-8 days in India and 1-3 weeks to the 25 other countries served; pocket and sleeve prints are delivered within India only.

create_shirt_from_photo

ChatGPT
Puts a user's photo on a garment and returns a new draft. style 'as-is' (default) removes the background and prints the photo directly: about 4-6 seconds, no AI image generation. style 'artistic' uses the photo as a seed for an AI illustration (for example 'Studio Ghibli style') and draws a new image with AI. Input: photo_url. In ChatGPT it is filled from the uploaded file; elsewhere pass the photo URL directly.

create_vector_design

ChatGPT
Creates a new draft from a structured layer stack of image, text and shape primitives, rendered on the server as SVG and then PNG. No AI image generation. This is the structured-JSON alternative to import_svg_design, which takes raw SVG markup. Coordinates are relative to the printable area (origin top-left). Layers stack in array order, layers[0] at the bottom; parts outside the area are clipped. get_canvas_state returns the printable bounds. Suited to typography, monograms, icons and geometric patterns.

generate_design

ChatGPT
Generates a new design with Crayonz's image model and creates a draft on the chosen garment. Takes plain product names ("t-shirt", "hoodie", "oversized tee") directly, so list_garments is not needed first. Returns draft_id, a signed preview URL (valid 15 minutes), a thumbnail and the price. Draws a new image with AI. Inputs: prompt (3-500 characters); garment_type and color (defaults can be changed later at no cost); placement (default front-center); num_images (default 2, up to 4); back_prompt for a back print; reference_image_urls (up to 3 style anchors); seed_image_url to print an existing image as-is with its background removed. The image is drawn server-side, so the host needs no image model of its own. An image that already exists can instead go to import_external_design. When generation continues in the background the response contains a job_id; get_generation or check_job returns the result. Follow-up tools: modify_design, add_text, add_shape, set_design_placement, set_purchase_options, create_checkout. A daily limit, when reached, returns an error that says so. Sexual, hateful or exploitative requests return content_blocked.

generate_lifestyle_shot

ChatGPT
Renders photorealistic images of a person wearing the draft's design in a chosen photographic vibe. The draft's design and garment mockup are used as references so the print matches. Inputs: draft_id, vibe, shot_count 1-6 (default 1). About 4-8 seconds per shot. Not counted as a design iteration. Vibes: streetwear, minimal, grit, y2k, kinfolk, editorial; scene vibes (hiphop, festival, academia, techwear, gym, retro_90s, beach, punk, outdoor_adventure, anime_otaku, skater, corporate_casual, college_campus, casual_kickback, date_night, gamer, music_scene, funny); sports fandoms (sports_fan, football_fan, cricket_fan, f1_fan, basketball_fan). recommend_photoshoot_vibes ranks vibes for a draft. Single-shot results are cached per design and vibe; changing the design (add_text, add_shape, modify_design, place_design, reset_layers) invalidates the cache.

get_canvas_state

ChatGPT
Returns the printable geometry for every view of a draft's garment, the current placement and the live mockup URL. Each view has printable_area_px (the bounding rectangle in canvas pixels, origin top-left) and design_areas (named regions such as Front or Left Pocket, with pixel rectangle and physical inches where the template sets them). Read-only. Useful for precise placement with place_design or create_vector_design.

get_design_suggestions

ChatGPT
Analyses a draft with a vision model and returns prioritised improvement ideas covering contrast, resolution, composition, colour harmony and printability, graded as error, warning or info. Read-only; involves no AI image generation. A suggestion can be applied with apply_suggestion or modify_design; size and position changes use set_design_placement.

get_draft

ChatGPT
Looks up one draft, or lists recent drafts. With draft_id: returns the full state, a 15-minute signed preview URL and the version history. Without draft_id: returns the user's 10 most recent drafts (id, garment, status, price, updated_at, public preview URL); response_format applies to this list. Read-only.

get_generation

ChatGPT
Returns the progress of a design that generate_design started in the background (stage, takes finished), and once it is done the finished drafts with their draft ids. Input: job_id from generate_design.

get_permanent_preview

ChatGPT
Saves a draft's current mockup(s) as permanent public URLs. Draft mockup URLs are signed and expire after 15 minutes; these do not, which suits links shared outside the chat such as a message, social post or email. views: ['front'] (default), ['back'] or both; a view with no mockup returns null. Repeat calls on an unchanged draft return the same URL; editing the draft produces a new snapshot.

import_external_design

ChatGPT
Use for a photo or image the person attached ("print my dog"): call it directly with the image, no list_garments first. Creates a draft from an image URL: an image generated with the host's own image tool, a Pinterest pin, a Supabase upload or a user attachment. It does not use Crayonz's own image generation. An image on its own has no garment, price or checkout link; this tool turns it into a draft that has them. For a text prompt with no image, generate_design draws the artwork. Expects artwork alone on a plain background (the motif, illustration or lettering by itself). An image that shows a garment (a flat-lay, a model wearing one, or a multi-panel mockup sheet) is printed as-is onto the garment. Resolution: a standard 1024px image is fine; create_checkout upscales to print resolution (2408x3408) at order time. Flow: the image is re-hosted, its background removed, the mockup composed and the draft preview returned. The draft then takes a size (set_purchase_options) and goes to create_checkout.

import_svg_design

ChatGPT
Creates a draft from SVG markup. Input: a complete SVG document with an explicit viewBox and a transparent background (no full-canvas background rect). It is rasterised to PNG at the printable area's native pixel resolution inside the Worker; background removal is skipped because SVG is transparent. Suited to typography, simple icons, monograms, geometric patterns and single-colour line art. No AI image generation. create_vector_design is the structured-layer alternative.

list_colors

ChatGPT
Lists every colour in the Crayonz catalog, deduplicated across garments, with the garments each colour is available on. It uses the same data as list_garments, organised by colour; clients that render the colour-palette widget show a hue-sorted grid. For garment-first browsing, see list_garments. Read-only.

list_community_designs

ChatGPT
Browses the Crayonz community feed, paginated. Combines studio generations (source='feed', anonymous) and marketplace listings (source='listing', creator-attributed). Inputs: limit 1-30 (default 12) and sort, 'newest' (default) or 'popular'. Returned item ids work with remix_listing and analyze_reference_image. Read-only.

list_design_versions

ChatGPT
Returns a draft's full version history as a structured list. Each entry has version_id, parent_version_id, stage (structural, polished, manual, ai_generated, rollback or import), view (front, back or both), a signed 15-minute preview URL and the prompt. Also returns the ancestry chain from the current version back to the first. version_ids are used with rollback_to_version. Read-only; involves no AI image generation.

list_fonts

ChatGPT
Lists the fonts the text renderer supports. Each entry has a status: bundled (embedded in the Worker, renders exactly) or fallback (add_text accepts the family but renders the bundled default, Inter Bold). Optional filters: category (sans, serif, display, script, monospace), weight_min and weight_max (100-900). Fonts are used through add_text's font parameter, or through a preset via add_text(preset=<slug>). Read-only.

list_garments

ChatGPT
Lists the Crayonz catalog: each garment with slug, name, sizes (apparel runs XS-7XL, headwear is one-size "NA"), INR price, colours (name, hex, availability, thumbnail) and supported placements. Optional garment_type filter. Use when the person asks what Crayonz prints on, or wants to choose a product. When they already named one ("a black t-shirt", "hoodie"), call generate_design directly: it takes the plain name, so this list is not needed first. Placement defaults to front-center. left-pocket and right-pocket are named from the wearer's side: left-pocket is the wearer's left chest, the viewer's right on the mockup. Read-only.

list_layers

ChatGPT
Lists every layer on a draft (logos, text, shapes, imported raster art) with layer_id, kind, view, region, z-order, scale and visibility. The layer_ids are used by update_layer, remove_layer and reorder_layers. Read-only.

list_my_logos

ChatGPT
Returns the user's saved logo library: every logo saved through upload_logo across sessions, with label and image_url. The image_url can be passed to compose_layered_design, place_logo or any tool that takes a logo URL. Read-only.

list_styles

ChatGPT
Lists the current user's saved style presets, newest first, with name, palette, font_style, notes and created_at. The names are used with apply_style. Read-only.

list_text_presets

ChatGPT
Lists typography presets for common merch text: retro and vaporwave headlines, jersey numbers, monograms, minimal wordmarks, batch years, farewell signatures and college names. Each preset bundles font, weight, colour, stroke and spacing under a slug used as add_text(preset=<slug>). Categories: batch-year, college, minimal, monogram, retro, signature, varsity; the category parameter narrows the list. Each preset has a preview_url (transparent PNG), which can be null on newly added presets; sample_text and notes describe the preset in that case. Read-only.

list_versions

ChatGPT
Returns the ordered version history of a draft, newest first: every generate_design, modify_design, add_text, add_shape and place_design change. Each entry has a signed 15-minute preview URL, the prompt snippet, the side touched (front or back) and a kind label. The response also shows which version reset_layers would revert to for a side. The history holds at most 20 entries per draft. Read-only; involves no AI image generation.

modify_design

ChatGPT
Edits a draft's artwork from a natural-language instruction (for example 'warmer colours', 'add a sun', 'replace the red triangle with a blue circle'). Re-runs the AI image edit and returns the same draft_id with a new preview and a version entry. Draws a new image with AI. Specific instructions give better results than vague ones. region_hint limits the edit to one part of the design ('his face') and leaves the rest unchanged. The first reference_image_urls entry is used as a style anchor. view='back' edits the back print, or draws one from the instruction when the draft has no back yet ('also put a phoenix on the back': +₹60 per piece). Position and size changes: set_design_placement. Garment colour: set_design_placement(garment_color). Text: add_text. Shapes: add_shape. None of the three uses AI image generation.

place_design

ChatGPT
Recomposes a draft's mockup with explicit pixel geometry, in one of two modes. Single: pass area, a rectangle on the printable canvas where the draft's existing design is placed; out-of-bounds values are clamped and the response includes clamped: true. Multi-layer: pass layers, an ordered array of image, text and shape layers with areas relative to the printable area (origin top-left, layers[0] at the bottom); layers clip at the edges. No AI image generation. get_canvas_state returns the printable bounds. For named slots such as 'left pocket' or 'back centre', set_design_placement is the matching tool.

polish_design

ChatGPT
Runs an AI aesthetic pass over an existing draft, such as a compose_layered_design result. Keeps the text and layout, upgrades typography, adds graphics to empty areas and adds atmospheric detail. Draws a new image with AI. view is 'front' or 'back' (default: the view with the most content); the other side is unchanged. Non-destructive: a new version is added and rollback_to_version can restore the earlier one.

publish_to_marketplace

ChatGPT
Lists a finished draft on the public Crayonz marketplace and returns listing_id and the public URL. The listing is visible to everyone. The creator earns their margin on each sale of this listing, and does not earn on remixes of it. Inputs: title (3-120 characters, required), description, tags (3-5 lowercase keywords), price (defaults to the draft's price; a higher value adds a creator margin), mrp (defaults to the garment's standard MRP). The draft must have status 'ready'. The listing then appears in list_community_designs.

recommend_photoshoot_vibes

ChatGPT
Ranks the lookbook photoshoot vibes that fit a draft's design, with reasons, using a vision model that looks at the artwork itself (a cricket graphic ranks cricket_fan, a racing car f1_fan). The result feeds generate_lifestyle_shot. Read-only, about 2 seconds, involves no AI image generation.

remix_listing

ChatGPT
Creates a new draft from a community feed or marketplace item. item_id comes from list_community_designs. Modes: 'reference' (default) anchors on the item's style plus the user's prompt, which is required; 'clone' reuses the artwork as-is on the chosen garment, colour and placement, with no AI generation and the prompt ignored. Defaults: tshirt, black, front-center. Remixes credit the original creator; the remixer does not earn on the original's sales.

remove_layer

ChatGPT
Removes one layer from a draft by layer_id. Other layers stay and the affected view is recomposed. If it was the only layer on its view, that view's mockup becomes null. The layer_id no longer exists afterwards; update_layer with visible=false hides a layer instead. No AI image generation.

reorder_layers

ChatGPT
Changes a layer's z_index. Lower values draw first (bottom of the stack) and higher values draw later (on top), for example to bring a text layer in front of a logo. No AI image generation.

reset_layers

ChatGPT
Reverts the front or back of a draft to the last generate_design or modify_design output on that side, removing every add_text, add_shape and place_design overlay added since. No AI image generation. view: front (default) or back. Errors with no_prior_version when there is nothing to revert to on that side.

rollback_to_version

ChatGPT
Undoes changes by moving a draft's current version pointer to an earlier version. Intermediate versions stay in the history, so the draft can move back and forth. The mirrored URL columns (current and back design and mockup URLs) are refreshed to match the target, and a new version with stage 'rollback' is recorded. No AI image generation. Version ids come from list_design_versions.

save_style

ChatGPT
Saves a named brand kit: a palette of up to 8 hex colours, a font style hint, an optional logo URL and optional notes (for example 'always centred'). Upserts by name: saving the same name again overwrites the earlier kit. No AI image generation. apply_style uses a saved kit on a draft.

search_logos

ChatGPT
Searches the curated institute logo library of 160+ approved logos (IITs, NITs, IIMs, AIIMS, NLUs, BITS, IISc, IISERs, IIITs). Each result has a slug, name, short name, city, brand colours and URLs for PNG, SVG and dark-mode PNG (variants can be null; png_url is the fallback). The free-text query matches full name, short name ('IITB') or city; an empty query with a series filter lists that series alphabetically. Slugs are used with place_logo and compose_layered_design. Read-only, about 200 ms.

set_design_placement

ChatGPT
Changes how an existing draft is made without redrawing the artwork, and returns the new mockup: position (front-center, back-center, left-pocket, right-pocket, left-sleeve, right-sleeve; pockets and sleeves are named from the wearer's side), scale 0.5-1.5, garment_color, or garment_type to put the same design on another garment (price and colour are recomputed; a size that does not exist on the new garment is cleared). Free and takes a few seconds. Artwork changes use modify_design. Pockets and sleeves are about 4x4 inches, so large artwork reads better front or back. Not every garment has every position (a zip hoodie has no front-center); list_garments shows each garment's positions.

set_purchase_options

ChatGPT
Sets size and/or quantity on a draft. create_checkout refuses a draft whose size is null, unless the garment is single-size. Metadata only: no AI image generation, no preview change. Sizes follow the garment (list_garments sizes). Apparel runs XS-7XL in Indian sizing, roughly US for adults (chest 40" is M, 44" is L, 48" is XL); 3XL and above add a small surcharge, quoted in the response. Caps, mugs, bottles and posters are single-size. Quantity 1-5; larger orders go through crayonz.ai/team-t-shirts. Both arguments are optional; omitted fields keep their current value.

update_layer

ChatGPT
Adjusts one layer on a draft without changing the others. At least one of scale, region, color, text or visible is required. Examples: scale=1.3 enlarges a layer, region="pocket-right" moves it, color="#FFD700" recolours it, text="BATCH 2028" edits its text, visible=false hides it. Only the affected view (front or back) is recomposed; the other side is unchanged. No AI image generation.

Capabilities

InteractiveWrites

Example Prompts

Click any prompt to copy it.

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/crayonz"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/crayonz)