MCP App Store
by KeenEthics
Photoshoot GPT: AI Fashion icon

Photoshoot GPT: AI Fashion MCP Server for ChatGPT

By Lonca Bilişim Teknolojileri A.Ş.

Overview

Turn your clothing images into AI-powered fashion photos and videos with Photoshoot GPT. Show your collection on models, reimagine existing fashion photos, and turn still images into videos for your online store, marketplace listings, and social media. Built for clothing brands, online sellers, and creative teams, Photoshoot GPT offers four ways to create content from your product and reference images. Product videos Turn a still product image into a video. Choose from available motion options to add movement to your photos and create content for product pages, collection launches, and social posts. Start with a photo you already have, without filming new footage. On-model product photos Create a new photo of a model wearing your clothing, starting with a product image. Choose a model, background, and pose to shape the presentation—from a clean catalog look to a styled scene for your next collection. Model replacement Create a new version of an existing fashion photo with a different model. Use a photo that already features your clothing and explore other model looks for the same outfit. Refresh your product photos or develop alternatives for a campaign without starting from a new shoot. Product replacement Use a reference fashion photo to imagine your own product in that setting. Replace the featured clothing with a piece from your collection, using the reference as a guide for the scene and composition. Explore a new presentation for your product or try different pieces within a similar visual concept. Describe what you want to create in your conversation, browse visual options, and choose the direction for your shoot. Your saved products, uploaded images, and generated photos and videos are also available in your Photoshoot GPT website account, so you can continue working across both. Supports English, Turkish, Spanish, and Portuguese. Requires a Photoshoot GPT account.

How to connect

  1. ChatGPT: Open the Photoshoot GPT: AI Fashion listing in the ChatGPT app directory and choose Connect. ChatGPT listing Photoshoot GPT: AI Fashion exposes 23 tools here.

Tools

approve_photoshoot_review

ChatGPT
App-only explicit summary approval. Requires the confirmation nonce delivered privately to the widget. Saves an approval intent for this exact revision and cost, with expiry; does not enqueue generation, reserve or spend credits. Reuse the same request key after an uncertain result.

ask_photoshoot_type

ChatGPT
Ask what to create using a native single-select form; waits for explicit user input, decline or cancel when the connection supports forms. Never an HTML/widget picker. Default to all four choices including product video; restrict media_type only for an explicit restriction. If status=question_required, no form was shown: display the question and ALL returned choices directly in the final chat message. Do not delegate this fallback to an asynchronous question tool. An accepted:true question acknowledgement is not the user answer. Never claim a selection UI was displayed from a question_required result. Wait for an explicit choice; do not choose a default. After any explicit type choice, including a numbered answer to question_required, call start_photoshoot with the chosen generation_type and a stable request key, then immediately render_photoshoot_catalog (products for create_on_model_photo/create_video, references for change_model/change_product). Existing account images come first; uploading is optional; declined/cancelled must stop without automatically asking again. This tool creates no draft or production and spends no credits.

browse_photoshoot_assets

ChatGPT
Browse real website models, product types, compatible backgrounds/poses, motion previews, or owned products/references for this draft. Choose a product before backgrounds, and a background before poses. IDs must come from these results. Use page_number for previous/next navigation and to keep the same page after saving a selection; returns page_number and total_pages. Legacy cursor requests remain supported but expire after a draft change; do not combine cursor and page_number. Does not save a selection.

confirm_photoshoot_action

ChatGPT
App-only explicit edit/retry confirmation. Requires the private nonce of the displayed action. Atomically records one operation and any website credit charge. Reuse the same request key on response loss; never submit a new action as an automatic retry.

generate_photoshoot

ChatGPT
App-only production action after summary approval. The widget can approve the summary and call this tool in sequence from one explicit user click showing total website credits and output count. Requires the private production nonce. Atomically reserves web credits and records one operation; never infer consent or invoke from conversation. Reuse the same request key on response loss; provider failures/timeouts follow the website refund policy; never resubmit an unconfirmed operation.

get_photoshoot_account

ChatGPT
Read the connected Photoshoot website account, latest subscription plan/status and current website credit balance. Use this to verify account connection. Does not create an API account, start generation, spend credits or change a plan.

get_photoshoot_operation

ChatGPT
Read the connected account’s durable operation after reconnect or production confirmation. Returns queued/running/completed/failed/partial/unknown, reserved/refunded website credits and safe final media links. Provider failures/timeouts are failed and refunded as on the website. Unknown denotes an unconfirmed local result record; refresh the same operation, never initiate a replacement generation or promise success.

get_photoshoot_results

ChatGPT
Read final image/video outputs for this account’s saved operation, including partial success and refunded credits. Use the original operation_id after reconnect. Refresh an unconfirmed local operation record instead of submitting again. Provider errors/timeouts follow the website refund policy. Use the returned flow IDs and permissions to prepare an explicit edit/retry action.

get_photoshoot_state

ChatGPT
Read this account’s saved photoshoot draft and current revision. Follow missing_fields and next_action. image_required means show the existing-image catalog (references when reference_image_id is missing, otherwise products), with uploading optional; it never means force an upload. review_required means a review is needed, not that production was authorized. No polling is needed while waiting for user selections.

import_photoshoot_image

ChatGPT
Save a user-provided hosted image as a new product or reference and select it in an existing Photoshoot draft. Use when the user explicitly asks to use that image in this draft; ask for clarification when intent is unclear. Read get_photoshoot_state for the current revision. The photo parameter accepts the real host-provided attachment metadata, even when no URL appears in the message. For local-only images without a hosted descriptor, use the catalog upload control. Product imports require a catalog product_type_id; references omit it. Accepts JPEG/PNG/WebP up to 10 MiB and 40 megapixels. Creates one new single-image source and replaces the source selection in this draft, which can discard dependent draft choices. Does not modify or delete existing products, references or generated media, or start generation. After saving, render the matching products/references catalog with focus_selected=true and wait for Confirm and continue. No credits are spent. Reuse request_key on retry; download_url may be refreshed for the same file_id.

list_photoshoot_products

ChatGPT
List only the connected website account’s products, including their thumbnails. Use a returned product_id with set_photoshoot_selection; this does not copy or edit the product.

prepare_photoshoot_action

ChatGPT
Prepare an edit or retry of one owned generation from an existing operation. Use real generation_id/flow_id from results. Image edit requires instruction and one completed image flow; website limit is two edit attempts, edits are free. Retry flow requires failed/not_started flow; retry_generation requires failed generation. Show resume/restart and quoted web credits in the action widget. This tool only saves a ten-minute summary; never performs an edit, retry or charge. Do not infer confirmation from conversation.

prepare_photoshoot_card_image

ChatGPT
App-only: authorize saving the image explicitly selected in the catalog upload control. Scoped to this draft revision, request key, image hash and size. No generation or credit spending.

prepare_photoshoot_local_image

ChatGPT
Legacy helper clients only; new clients use the catalog upload control. Legacy exact-byte preparation for older local clients. Prefer prepare_photoshoot_local_import and the local import_photoshoot_local_image tool without shell commands. Explicit user intent to use the attachment authorizes saving. For legacy clients only: run the bundled local-image.mjs describe helper on the exact attachment path, then pass its file_sha256, byte_size, file_name and mime_type here with the draft revision, purpose and real product type ID. Returns a ten-minute upload authorization bound to those exact bytes and this draft. Use the bundled upload helper with the returned JSON on stdin; it sends the file directly to Photoshoot and returns the saved draft. Never paste base64, invent a download URL, read OAuth credentials or fall back to browser uploads. Reuse the same request_key on lost responses; if authorization expires, prepare again with that same key. No production, image analysis or credit spending.

prepare_photoshoot_local_import

ChatGPT
Legacy helper clients only; new clients use the catalog upload control. Authorize saving one user-attached LOCAL Codex image to this draft. Explicit intent to use the image authorizes saving without another question; ask only if intent is unclear. First browse product_types for a real product_type_id when purpose is product. Pass the returned authorization unchanged to the LOCAL import_photoshoot_local_image tool with the exact attachment path and conversation language. That tool reads metadata and uploads bytes internally; never run node/shell helpers or create temporary token files. No production, analysis or credit spending. Reuse the same request_key, revision and intent after a lost response. After confirmed product save reopen render_photoshoot_catalog with kind products and focus_selected true; wait for Confirm and continue.

prepare_photoshoot_review

ChatGPT
Prepare a ten-minute shooting summary for the current draft revision, using authoritative website credit costs and current website balance. Use a stable request_key for retries. Then render_photoshoot_review so the user can review selections and explicitly approve in the widget. Never infer approval from conversation or catalog clicks. Saves only a summary; does not generate media, reserve or spend credits.

render_photoshoot_action

ChatGPT
Show the exact edit instruction, target image, retry mode and web credit cost for separate explicit user confirmation. Stale/expired summaries must be prepared again. After confirmation the widget sends a chat message with photoshoot_operation in the updated context. That edit/retry operation is already admitted: wait_photoshoot_operation follows its saved ID until settled, then render_photoshoot_results opens a separate results card. Never confirm or retry it again while tracking. This read never authorizes execution.

render_photoshoot_catalog

ChatGPT
Open a visual picker for a real catalog or the account’s existing products/references. For an image the user explicitly wants to add to this draft, save the source using import_photoshoot_image with a hosted descriptor or the catalog upload control for a local-only image. If importing is unavailable, explain the limitation. After saving a new product/reference, reopen this picker with the matching kind and focus_selected true so the saved item is visible and selected. Wait for Confirm and continue before advancing. Selection clicks save only the draft. The user then clicks Confirm and continue to send a chat message with the saved draft reference in readable model context. Continue from that saved session/revision: read state and show the next missing catalog, or prepare/render the shooting review when ready. Do not ask the user to type that they selected. This confirms catalog choices only, never authorizes generation or spending credits. Product types are informational for the next attachment import.

render_photoshoot_results

ChatGPT
Open a separate result card for an already-started operation. After the confirmation message, wait_photoshoot_operation follows the saved operation; when finished call this tool with the same operation_id and the conversation language. Displays final image/video previews, partial failures and website credit refunds. Never re-render the original summary as the progress/result screen and never generate or retry from this tool. If opened while queued/running, this separate result card reads status automatically until settled.

render_photoshoot_review

ChatGPT
Show a prepared shooting summary with selected images, aspect ratio, output count, total website credit cost and current plan/balance. With production permission, one explicit button shows the total website credits and output count, validates and approves this summary, then starts production. Never infer that click from conversation. Refresh summary remains a separate action. After admission the widget sends a chat confirmation message. If an operation_id is already present, follow it with wait_photoshoot_operation and render_photoshoot_results as a separate card; otherwise refresh a stale or expired summary.

set_photoshoot_selection

ChatGPT
Save explicit user selections in the current photoshoot draft. Use real owned/catalog IDs, the expected revision and a stable request key. Saving never approves a generation or spends credits. This overwrites draft choices and may discard dependent draft selections when the product or background changes; those choices must be selected again. Existing products, references and generated media are not modified or deleted. Reload on revision conflict.

start_photoshoot

ChatGPT
Open an empty Photoshoot draft for one of the four product types. This saves a draft only; never generates media or spends credits. Use a stable request_key for retries. Immediately follow with render_photoshoot_catalog: products for create_on_model_photo/create_video, references for change_model/change_product. Do not stop to request an upload; existing images are the default and new uploads are optional.

wait_photoshoot_operation

ChatGPT
Follow an already-started Photoshoot operation for up to 25 seconds. Does not create, retry, reserve credits or start a replacement production. After the widget confirmation message, use photoshoot_operation.operation_id from the updated conversation context. If queued/running, call this tool again with the same ID until it finishes; do not ask the user to refresh. When completed/partial/failed/unknown, open render_photoshoot_results in a NEW result card. For unknown report the unconfirmed record without promising success or starting another generation. Stop on cancellation, authentication error or user request.

Capabilities

WritesInteractive

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/photoshoot-gpt-ai-fashion"><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/photoshoot-gpt-ai-fashion)