Model Context Protocol
Connect Claude, ChatGPT or Cursor once. Your AI chat then runs your templates, workflows and apps. It reads your brand rules and shows what a run costs first.
75 MCP tools behind one URL. Reading your workspace charges no credits.
Product photo
You will paste this into your AI client.
https://mcp.designerbox.ai/apiIn Claude Desktop or Claude.ai, open Settings then Connectors, click Add custom connector, name it DesignerBox and paste the URL.
In ChatGPT, open Settings then Connectors, add a connector in developer mode and paste the URL, then approve the DesignerBox sign-in.
Add DesignerBox to your Cursor MCP config, then approve the sign-in when your editor asks.
{ "mcpServers": { "designerbox": { "url": "https://mcp.designerbox.ai/api", "transport": "http" } } }Point MCP Inspector at the URL, pick Streamable HTTP, and approve the sign-in in the browser window.
For Claude Code, a script or a build server, send the key as a bearer token. An API key is on the Ultra plan.
POST https://mcp.designerbox.ai/api Authorization: Bearer apikey_xxx Content-Type: application/json
Approve the DesignerBox sign-in once. Your client holds a scoped token for 30 days. Then you ask it for a job in plain words.
Send the key with every call. The same 75 MCP tools answer, with no browser step.
Prefer a walk-through with every client? Connect page
How it works
The same jobs you run in DesignerBox, asked for in plain words.
A workflow you built, with your brand and your steps. Every step returns its own result.
Your chat repeats one job for every row you paste. A batch in DesignerBox does the same for a spreadsheet.
Step one is a template. Step four is the whole catalogue.
Jobs
Copy a line and paste it in your chat. Each card names the MCP tools your chat calls.
Give it a product photo. It returns a PNG with no background.
Using DesignerBox, remove the background from this product photo and give me a transparent PNG.
remove_background
Give it a product photo. It returns the same product from a new angle, in the same light.
Using DesignerBox, make a three-quarter view of this product from this photo. Hold the light and the framing.
photo_angles
Give it several product photos. It returns one flat lay composition.
Using DesignerBox, build a flat lay from these three product photos on warm linen, seen from above.
flat_lay
Give it a product photo. It returns an ad picture you can run.
Using DesignerBox, make a product ad from this photo for paid social, portrait, with room for a headline.
product_ad
Give it a garment photo and a model photo. The face, the body and the pose stay.
Using DesignerBox, put this garment on this model photo. Keep her face, her pose and the background.
virtual_try_on
Give it an on-model photo. It returns a fashion clip that keeps the outfit.
Using DesignerBox, make a fashion video from this on-model photo: one slow turn, the outfit unchanged.
fashion_video
Give it a portrait. It returns clean skin with the pose and the background kept.
Using DesignerBox, retouch the skin in this portrait. Keep the pose, the light and the background.
skin_retouch
Give it a brief. It returns nine poses of one person.
Using DesignerBox, create an avatar of a woman in an ivory linen shirt, daylight, looking at the camera.
create_avatar
Give it a face and a script, or your own audio. It returns a lipsync clip.
Using DesignerBox, make a talking avatar from this face reading this script, in a warm female voice.
talking_avatar
Give it a product photo. It returns a short clip with the move you describe.
Using DesignerBox, animate this product photo into an eight second clip, a slow push in, 16:9.
generate_video
Give it your clips. It returns one MP4 with titles, sound and a look.
Using DesignerBox, build a video project from these three clips with a crossfade, a title card and music, then make the MP4.
create_video_project
Give it a flat photo. It returns the same shot under the light you describe.
Using DesignerBox, relight this photo with soft studio light from the left, one stop brighter.
spotlight
Give it a picture and one instruction. The detail and the resolution stay.
Using DesignerBox, change the label on this jar to matte black. Leave everything else the same.
edit_imageA run charges credits. Your client asks you to confirm the call first.
Workflows
A workflow holds your brand, your products and the steps in order. Your chat can run one, build one, or run the app version for a colleague.
Your chat lists your workflows and runs the one you name. Each step returns its own result, so you can see where a run went wrong.
Using DesignerBox, list my workflows, then run my product set workflow on this photo and show me each step.
Describe the steps once. Your chat saves the workflow in your workspace, where you or a colleague can run it again.
Using DesignerBox, build a workflow that clears the background, relights the product and makes three scenes, then save it as Product set.
An app is a workflow behind a form. Your chat fills the fields, starts the run and follows it until the results arrive.
Using DesignerBox, run my on-model app with this model photo and these two garment photos, then tell me when it is finished.
Inputs
Your chat opens a file picker. The photo goes straight to Assets and is ready to use.
Every design, asset and folder is readable by name, so a run can start from work you already have.
One search covers Unsplash, Pexels and Pixabay, and returns them together.
Working in Figma? The DesignerBox plugin runs the same jobs beside your file. Figma plugin
Brand
Your active brand profile loads at the start of a chat: the voice, the themes, the industry. Nobody has to describe the brand again.
Your logos, fonts and palette sit in Assets, readable by the same chat. A workflow reads the brand record before every run.
Hand your chat a style guide and it can create a brand profile, make it the default, or update the default one in your workspace.
Cost and control
Billing is the loudest complaint in this category. Here is what a chat can and cannot spend.
Reading your designs, brands, models and plan charges no credits.
Your chat can read what each model costs and how many credits you have left, before it starts.
Every finished run reports the credits it charged, in the same answer as the result.
A repeated call within five minutes returns the first result instead of charging twice.
Stop an avatar or a video job while it runs and the credits are refunded.
A template runs by name. Its prompt is loaded on our server and never sent to the chat.
Reference
One URL, 75 MCP tools. Reading charges no credits. A run charges credits and your client confirms the call first.
list_designsget_designdelete_designsearch_designslist_folderslist_assetsget_assetdelete_assetlist_gallerieslist_whiteboardsget_whiteboardlist_conversationsget_conversationgenerate_imageedit_imageapply_image_templategenerate_designgenerate_cinema_studioremove_backgroundphoto_anglesspotlightflat_layproduct_advirtual_try_onskin_retouchgenerate_videoget_video_statuslist_video_taskscancel_video_taskfashion_videostyled_scenecreate_video_projectrender_video_projectget_video_render_statusdelete_video_projectrun_pipelinelist_pipelinesget_pipelinecreate_pipelineupdate_pipelinedelete_pipelinerun_appget_app_runlist_community_workflowsget_community_workflowcreate_designlist_brand_profilesget_brand_profileupdate_brand_profilecreate_brand_profileset_default_brand_profilelist_brand_assetsget_brand_assetcreate_avatarlist_avatarsget_avatarget_avatar_statuscancel_avatar_tasktalking_avatarget_talking_avatar_statuslist_image_templatesget_image_templatelist_photo_packsget_photo_packsearch_stock_photoslist_skillslist_image_modelslist_video_modelsget_planget_plan_limitslist_teamsset_current_teamsearch_toolsimport_chat_imageupload_assetFive read-only records your client reads when the conversation opens, with no prompt to approve.
Your chat reads what each model costs, then picks one for the job. You can also name the model yourself.
Plans
Reading your workspace works on every plan. Credits pay for the runs. The price and the credits of each plan are on the pricing page.
One run at a time, on the sample products.
No DesignerBox mark on your pictures. One person.
Your own product photos, your own brief and a commercial licence.
Video, virtual try-on, upscaling and both editors.
Five people, shared brand kits, white label and the API.
Questions
Short answers about connecting, credits and what a chat may do.
DesignerBox runs a Model Context Protocol server at one URL. You add that URL to Claude, ChatGPT or Cursor and sign in once. Your AI chat then has 75 MCP tools. It can read your workspace, run your templates, workflows and apps, and make pictures and video in your brand. There is nothing to install.
Claude Desktop, Claude.ai, Claude Code, Cursor, ChatGPT and MCP Inspector, and any other client that speaks MCP over HTTP.
Product photos, on-model shots, flat lays, ads, avatars in nine poses, talking presenters, product and fashion video, and one finished MP4 from the video timeline. It can also edit a picture you already have and hold the resolution.
Reading charges nothing. A run charges credits, and your client asks you to confirm the call first. Every finished run reports the credits it charged. A repeated call within five minutes returns the first result instead of charging twice, and a cancelled avatar or video job is refunded.
Not for an AI client. Claude, ChatGPT and Cursor sign in with OAuth. For Claude Code, a script or a build server, make a key in the app under Account then API Keys and send it as a bearer token. An API key is on the Ultra plan.
Reading your workspace works on every plan, including Free. Runs use credits, and each plan has a monthly allowance. Video and virtual try-on start on Premium. A commercial licence starts on Pro. The plan page lists what each plan holds.
A picture comes back in the same call, and the time depends on the model. Video runs in the background: your chat sends the job, checks its status, then gives you the file. A video timeline takes one to five minutes to finish.
Yes. Your chat repeats the job for every row you paste, one row at a time, and tells you what each row cost. For a spreadsheet of products, a batch inside DesignerBox runs one workflow over every row and shows the results in one view.
Your active brand profile loads when the conversation opens, and your logos, fonts and palette are readable from Assets. A workflow reads the brand record before each run, so the brand is a record, not something a person has to remember.
Sign-in is OAuth, so your client holds a scoped token and never your password. The token lasts 30 days and you can revoke it in your account. A template's prompt stays on our server. Only the team you are signed in to is visible.
Connect Claude, ChatGPT or Cursor in three steps. Reading charges no credits, so you can look before you run anything.