configure_unify_image
ChatGPTFor an explicit slash command or named preset, resolve it with get_unify_preset_instructions first, then pass its native image preset ID in preset. For ordinary requests, use first when the user asks to make, create, generate or edit an image, photo, illustration or logo without finalized settings. Open the image creation card immediately with their idea; missing model or settings are not a blocker. Defaults to Nano Banana 2. Pass actual attached reference_images for editing or visual guidance. The card offers Nano Banana 2, Nano Banana Pro, Seedream 5 Lite/Pro and Grok Imagine, with their resolution and aspect settings, uploads and @Image mentions. Opening it is free and requires no login. The card quotes the exact Unify credit cost; clicking Generate approves and submits it once. The same card shows Generating, then images and download buttons. Do not ask for a follow-up chat confirmation, send a message to show the result, or submit again. For explicit complete quote requests, existing approvals or status checks, use the corresponding tool directly.
configure_unify_video
ChatGPTFor an explicit slash command or named preset, resolve it with get_unify_preset_instructions first, then pass its native video preset ID in preset. For ordinary requests, use this first when the user asks Unify to make, create, or generate a video, clip, or animation and has not finalized generation settings. Examples: "lets create a video of a dog", "make a video", "animate this logo", or "turn this image into a video". Open the video creation card immediately with their idea in prompt; the user does not need to ask for settings, know tool names, or provide a model, duration, resolution, or aspect ratio. Available models: Seedance 2.5, Kling 3.0 Motion, Wan 3.0, Gemini Omni 1.1 Flash and FLUX 3.0. Kling Motion requires one character image and one driving video. Omit unspecified options: the card supplies supported defaults and lets the user adjust them. Set reference_mode to chat-images when attached images should guide the video. Opening the card is free and requires no account connection. It does not generate, import attachments, or spend credits. Pass all intended attached image, video and audio file objects when opening the card. The card quotes directly and displays a Generate button with the exact cost; that click approves and submits the generation. Do not ask for another chat confirmation or submit it again. Legacy cards may send a quote request in chat; quote those settings and obtain approval there. Do not reopen the card for an existing quote, an approval, a status check, or a request only to write a prompt. For an explicit quote request with complete settings, quote directly.
generate_unify_media
ChatGPTSpend the connected user's Unify credits to execute an approved quote. Set max_credits to the user's approved credit limit. Use a unique idempotency_key per intended generation, and reuse the same key and quote after a timeout or retry. The server enforces the quote, account ownership and credit limit. For a generation submitted by the model, immediately call render_unify_generation with the returned generation ID. Generations submitted from the Unify card already show progress and the result in that card; no follow-up message or separate player is needed. Never start another paid job to check progress.
get_unify_account
ChatGPTGet the connected Unify account's available credit balance. Reading the balance is free. Generation spends this same account's Unify credits.
get_unify_generation
ChatGPTCheck an existing generation belonging to the connected Unify account. Returns status, credit information and output URLs when ready. This never starts a new generation. Use the generation ID returned by generate_unify_media; wait between checks while processing.
get_unify_model
ChatGPTUse this for details about one Unify model after searching the catalog. Pass the exact slug returned by search_unify_models. Returns its public features and links, not a generation result.
get_unify_preset_instructions
ChatGPTResolve commands for the selected Unify app. Unique shortcuts: /unify-hero-shot, /unify-product-spin, /unify-marketing-resize-ads, /unify-ugc-flow, /unify-exploded-view. Also accepts short aliases such as /hero-shot when the user selects or explicitly names Unify. A bare shared command without provider context is not an instruction to prefer Unify over another selected app. Keep subsequent media calls on this Unify connection; do not load another provider's preset skill or generation tools as a fallback. Pass the exact token, not the brief. Returns the recipe, required inputs, supported defaults and workflow or editor requirements. Omit preset to list commands. /unify-product, /unify-motion, /unify-effects and /unify-marketing browse without generation. Ordinary image/video requests use configure_unify_image/video directly. Quoted, discussed or negated tokens are not invocations. Resolution is free and does not generate or install anything.
get_unify_presets
ChatGPTBrowse or search Unify product-image, motion, effects, marketing, UGC, faceless and external-editor commands without generating. Returns real command IDs and requirements, not provider generation IDs. Resolve a selected ID with get_unify_preset_instructions before using it. Paginate only when requested. This is Unify's independently implemented catalog; it does not access Higgsfield jobs, credits or proprietary galleries.
list_unify_generation_models
ChatGPTList the image and video models available to generate directly through this connection, including accepted settings. Use these exact model identifiers when requesting a quote. The public website catalog includes additional models that are not supported here yet.
list_unify_recent_generations
ChatGPTList the connected user's 12 most recent submitted image and video generations for the Unify app home. This never creates or charges for a new generation.
prepare_unify_reference_upload
ChatGPTAuthorize one local reference upload from the video card. No generation or credit charge. The server validates file bytes before quoting.
quote_unify_generation
ChatGPTPrepare an expiring, account-bound price quote for the exact prompt, model, settings and optional reference files. Pass actual host file objects in reference_images, and supported reference_videos/reference_audios according to the selected model limits. Never invent download URLs or omit a selected file. The quote securely saves all references, verifies input video duration and includes its cost. This does not generate media or spend credits. Present the quoted cost to the user and obtain approval, unless their existing budget covers it. Check list_unify_generation_models for settings and reference limits.
render_unify_generation
ChatGPTShow an existing generation as an inline image or video player in ChatGPT. Call immediately after generate_unify_media, including while queued or processing: the card checks status automatically while open. Also use to reopen an earlier result. Pass only the returned generation ID. This checks ownership and never starts or charges for a new generation. Clients without UI can use the returned output URLs.
search_unify_models
ChatGPTUse this when the user wants to discover or compare models in Unify's public catalog. Search with short keywords, such as Kling, Google, or motion. All keywords must match. Omit query to browse; filter by category and paginate with offset. Returns website descriptions and links.
