---

name: product-photoshoot description: Create faithful studio, lifestyle, and on-model product photography through the GooseWorks Product Photos workflow, then approve selected results for reuse in future creative work.

Product Photoshoot

Turn real catalog images into publish-ready product photography while preserving silhouette, materials, logo, packaging, and colorway. The GooseWorks backend uses the same generation, fidelity review, and retry pipeline as its Product Photos studio.

Prerequisite

This workflow requires the GooseWorks MCP tools. If they are unavailable, tell the user how to install the GooseWorks MCP connection and stop before generation.

Workflow

1. Resolve the brand with list_ad_brands and the product with list_brand_products.
2. If the product is missing, use import_product with a product URL, Shopify store, or public image URL. Poll get_product_import; do not re-submit an in-progress import.
3. Clarify the intended image: studio, lifestyle, on-model, close-up, setting, aspect, and count. Do not invent product attributes.
4. Call estimate_product_photos and show the user the credit estimate. Confirm count and quality before spending.
5. Call generate_product_photos with the chosen product, category, controls, count, quality, and optional prompt. Human model imagery requires the user's rights attestation.
6. Poll get_product_photo_generation until complete, partial_failure, or failed. Do not submit a duplicate while it is running.
7. Show every result and status. Let the user choose the keepers; use approve_product_photo only for selected results and archive_product_photo for rejected ones.

Tool map

• Brand and catalog: list_ad_brands, list_brand_products, import_product, get_product_import
• Cost and generation: estimate_product_photos, generate_product_photos, get_product_photo_generation
• Results: list_product_photos, approve_product_photo, archive_product_photo

Rules

• Ask before spending credits.
• Approved photos become reusable brand creative inputs; unapproved photos do not.
• A fidelity-flagged output may be shown for review but must not be described as approved.
• Never imply that a generated person is a real customer or spokesperson.