analyze_amazon_listing_product
ChatGPTRun the same combined Step 1 used by the GreenOnion product-listing UI. It produces product analysis and the coordinated nine-slot feature assignment together. Present one combined report in this order: Product Analysis (including target audience and brand tone), Design System (font and primary/secondary colors), then all nine Feature Assignment slots with image type, the complete itemized feature list, and enabled state. The successful result includes review.display_mode=verbatim; return review.markdown exactly as the entire next response, then stop for the user's changes.
attach_amazon_product_photos
ChatGPTAttach one to five product photos supplied by ChatGPT to an existing, owned Amazon listing draft, then continue the requested listing workflow.
confirm_amazon_listing_analysis_and_feature_assignment
ChatGPTPersist the user's explicit acceptance of the complete product-analysis and feature-assignment version that was just shown. expected_version must match the current version. After confirmation, ask the single optional creative-direction question described by the server instructions. Save only the specifics the user supplies; omitted slots and fields remain AI choices.
confirm_amazon_listing_image_prompts
ChatGPTLegacy recovery tool for an image-prompt version that was manually revised. It is not part of the standard workflow because newly generated prompts are confirmed internally.
create_amazon_listing_draft
ChatGPTTemporary compatibility endpoint for conversations holding an older cached GreenOnion schema. New conversations must use create_amazon_listing_draft_v2. This tool accepts both previously published draft input formats, preserves supplied product facts and preferences, defaults omitted fields to AI, and normalizes output quality to 2k.
create_amazon_listing_draft_v2
ChatGPTCreate an account-owned GreenOnion draft with exactly nine Amazon listing slots. Use facts and preferences the user already supplied. Encode every unspecified Product Analysis or Design System value as source=ai_pick without asking the user separately, and encode absent additional product facts as source=none. The workflow always starts with all nine slots at the Amazon app's fixed 2k quality.
download_amazon_listing
ChatGPTCreate a protected 10-minute ZIP download for the selected images from a completed, account-owned Amazon listing job. Keep the saved GreenOnion project unchanged and offer the download after generation alongside the project link.
estimate_amazon_listing_cost
ChatGPTUse only when the user explicitly asks about credit usage or cost. Calculate the GreenOnion credit requirement without creating or changing anything.
generate_amazon_listing_image_prompts
ChatGPTLegacy recovery tool for conversations that already began the older two-call workflow. Do not use it for a standard workflow. start_amazon_listing_generation now prepares prompts inside the durable background job. This legacy operation uses the authenticated GreenOnion account's credits when it prepares prompts.
get_amazon_image_job_result
ChatGPTGet the protected GreenOnion project result for an account-owned job. This returns no original image or storage URLs.
get_amazon_image_job_status
ChatGPTRead the current stage and progress for an account-owned Amazon generation job. This is inexpensive and safe for a status widget to call with backoff.
get_amazon_listing_draft
ChatGPTRead one owned GreenOnion product-listing draft by ID so a user can resume or review an interrupted workflow without creating a new draft. If a review is returned, display review.markdown verbatim rather than reconstructing the analysis or slots.
get_amazon_listing_image_prompts
ChatGPTUse this when the user wants the listing images created here in ChatGPT instead of by GreenOnion. Call it after the analysis and plan are confirmed and the slot requirements are saved. It returns one ready-to-use image prompt per enabled slot. If the prompts have not been prepared yet, it prepares them first, which uses 1 GreenOnion credit per enabled slot (tell the user before calling); asking again later is free. Then generate one image per prompt with the user's product photo attached, in slot order, and save the finished set with save_amazon_listing_images (1 credit per image). Do not call generate_amazon_listing_image_prompts first. When you tell the user the cost, use prompt_credits (what the prompts cost in total), not credits_charged (only what this call charged). Send a new request_id per draft and reuse it when retrying.
list_amazon_listing_drafts
ChatGPTList recent owned GreenOnion product-listing drafts so a user can resume an interrupted workflow in a later conversation.
list_recent_amazon_image_jobs
ChatGPTList recent owned Amazon image generation jobs so a user can recover progress or completed results after conversation context is lost.
revise_amazon_listing_analysis_and_feature_assignment
ChatGPTApply user-requested edits after presenting the complete combined analysis. Any Product Analysis value, Design System value, or slot can be changed. Use product_analysis for typed field patches and slot_updates for only the slots and fields the user named. The server preserves untouched slots, and disabled slots remain in the nine-slot plan. After saving, return review.markdown exactly as the entire next response and stop for another review.
revise_amazon_listing_image_prompts
ChatGPTLegacy recovery tool for an older workflow in which image prompts were shown. Do not use it in the standard workflow. expected_version must match the older displayed version; prompt_updates applies requested changes to enabled slots only.
save_amazon_listing_images
ChatGPTSave listing images that you generated yourself in this chat into the user's GreenOnion project for this draft, so they appear in the GreenOnion web app with the project, can be edited, and are included in download_amazon_listing. Pass each finished image in the parameter for its slot: image_1 is slot 1 (the main image), image_2 is slot 2, and so on. Save all finished images of the set in one call, so they download as one ZIP. Use this only for finished listing images, not for the seller's product photos (use attach_amazon_product_photos for those). Saving uses 1 GreenOnion credit per image, the same as an image GreenOnion generates. Tell the user the credit cost before saving. Send a new request_id for each new save, and the same request_id when retrying the same save, so nothing is saved or charged twice.
save_amazon_listing_slot_requirements
ChatGPTSave only the specific image requests the user provided after confirming the analysis and feature assignment. For a mentioned slot, independently encode design and text_copy as source=user or source=ai_pick. Omit unmentioned and disabled slots; the server defaults every omitted enabled slot and field to AI. Use an empty requirements list when the user has no specific requests.
set_amazon_listing_active_slots
ChatGPTUse only when the user explicitly asks to generate a subset of the nine proposed images. Pass every slot number the user wants to keep; at least one is required. Return the complete updated report using review.markdown verbatim, including disabled slots.
start_amazon_listing_generation
ChatGPTAfter analysis approval and slot requirements are saved, queue one durable background job that prepares internal prompts, generates enabled images, and saves the project. This operation reserves and uses the authenticated GreenOnion account's credits shown in the approved review. Return immediately and reuse the request ID for transport retries.
start_listing_image_workflow
ChatGPTUse this when a user wants to create, improve, or plan one or more Amazon listing images or product infographics. It supports main images, feature and benefit infographics, comparisons, dimensions, close-up details, and lifestyle images—from one requested image to a coordinated set of up to nine—even when the user does not mention GreenOnion or has not provided product photos yet. Do not imply Amazon affiliation or platform-specific compliance; do not use it for general social posts, display ads, or non-product artwork.