ApparelHub

Set channel-defined listing attributes on ONE product (Material, Style, Washing Instructions and similar). Call describe_listing_attributes first to learn the field keys and their allowed values. A PARTIAL WRITE SUCCEEDS. Send four values with one bad and the three good ones are stored while the bad one is reported — you do not have to get them all right at once. A value the channel refuses comes back in `rejected` with a machine-readable `reason` and the allowed values echoed, so you can correct it in one more turn rather than guessing. Rejections are never dropped silently. ⛔ NEVER INVENT A VALUE. Relay what the merchant told you. If you cannot get a value from them, leave it UNSET and say so — an unset field is honest, an invented one is not. Do not infer it from the product type, do not copy it from another shop, and do not pick the nearest allowed value because it looks close. 📏 SIZE CHART. US apparel is graded down without one. A chart is normally rendered automatically from the fulfillment provider's real measurements, so most listings need nothing. When one IS flagged, prefer `size_chart_measurements` (an object — call import_size_measurements to fill it from the provider) over `size_chart_template_id`: the template id can only come from a human in the channel's own admin, because the channel publishes no way to list, verify or correct one. ⛔ NEVER INVENT MEASUREMENTS. They are what a buyer reads before choosing a size. Do not derive a table from the garment type, do not copy one from a similar product, and do not fill a gap with a plausible number. A malformed table is refused whole, with a reason — nothing is half-applied. A missing cell is fine and renders blank; an invented one means somebody receives a garment that does not fit. Setting a value does NOT change the live listing on its own — the channel is updated on the next sync. Pass `sync: true` to push it immediately, or run sync_to_channel afterwards. [#ef218b]

Change the recorded payment method on an order that already has a payment recorded (e.g. correct "stripe" to "sales_channel"). This is a bookkeeping label change; it does not move any money. [#588a7f]

Set each variant's price to hit a target profit margin off its OWN cost: price = cost / (1 - margin). Reads per-variant production cost (populated after the fulfillment sync), applies a per-variant price, then re-syncs connected channels. Use this instead of one flat price when costs tier by size (larger sizes cost more, so a single price gives a different margin per size — and can go negative on the biggest). Requires store_uuid (cost lives on the store-products list, not product detail). [#54cd6b]

Set an existing product's listing images: attach an uploaded photo or a generated lifestyle shot, reorder them, and choose the cover. Use this AFTER the product exists — create_product / ship_product pick the initial mockup themselves. ⚠️ THE LIST REPLACES, IT DOES NOT MERGE. What you send becomes the whole gallery, in the order given. To add one image, READ the current list first and send it back with the new entry in it — sending the new entry alone deletes every other image. Pass `images: null` to reset the gallery back to the product's provider mockups. ⚠️ ORDER IS FUNCTIONAL, NOT COSMETIC. Channels cap how many images a listing may carry and TRUNCATE IN GALLERY ORDER, so position decides what actually ships: TikTok Shop takes 9, Wix 15, Shopify and WooCommerce are unlimited. On a capped channel an image in position 10 is not a lower-priority image, it is an absent one. Put the images that must survive first. The platform stores at most 20. Each entry carries provenance. `source` says where the file came from (mockup / upload / ai_mockup / print_file / unknown). `ai_generated` is SEPARATE and tri-state on purpose: an uploaded photo may itself have been AI-generated and the platform cannot detect that, so only you can say. Set it truthfully — true, false, or leave it unset when you genuinely do not know. Do not guess it from `source`. `cover` sets the display image independently of order, so the cover need not be first. A cover that is not in the gallery is added to it. Replace the gallery without naming a cover and the cover follows to the new first image. CONCURRENCY: this reads the product first and passes its version back with the write, so a change someone else made in between is REFUSED rather than silently overwritten. On a conflict the tool re-reads and returns `conflict: true` with the current images — it does NOT retry, because the list you built was based on a gallery that no longer exists. Rebuild from `current_images` and call again. [#781240]

End-to-end pipeline in ONE call: take a design, generate + verify a mockup (one per imported color, so every color variant has a matching mockup), create the product with the correct field names, add all variants, associate with a store, sync to fulfillment, then (optionally) sync to sales channels as DRAFT. Handles EMBROIDERY garments automatically WHEN the garment is actually embroidered: routes the design to the real embroidery placement and attaches thread colors (derived from the design, or pass thread_colors). Headwear is NOT inherently embroidered -- some providers carry printed (DTF) caps that take photoreal art as-is, so check `accepts_photoreal` on the garment rather than assuming, and call find_garments before concluding a design cannot go on a hat. Face goods (canvas, posters, backpacks, bags, socks, towels, blankets, pillows, cases...) default to print_style "fill": the design is recomposed onto an aesthetically matching background and printed edge-to-edge, so no green-screen background or contrasting borders reach the product. Enforces pricing floors and guards the AQUA-vs-Navy variant trap. Streams progress. PREFER this over chaining create_product + add_variants + sync_to_fulfillment + sync_to_channel yourself — especially for AUTOMATED or SCHEDULED runs — because it guarantees the correct order (store association + fulfillment sync BEFORE any channel sync). Use the split primitives only when you deliberately need a partial/interactive flow. [#232f74]

Begin a browser-based connection (Printful, Shopify, TikTok Shop, Fourthwall). Shopify additionally requires shop_url, the merchant myshopify domain — ask for it before calling. Returns an authorization URL to give the user. THE CONNECTION IS NOT FINISHED WHEN THIS RETURNS. You must keep polling check_connection_status (passing the same provider_uuid) until it reports connected, then tell the user. The browser tab where they authorize is NOT this conversation and cannot report back to you, so polling is the only way you or they will learn it worked. Poll every few seconds, up to about two minutes, and if it has not landed by then ask whether they finished authorizing rather than giving up silently. If they need to create an upstream account first, let them, then call this again for a fresh link. [#6d4006]

Manually submit an order to its fulfillment provider (Printful/Printify) as a DRAFT. For sales-channel orders this auto-fetches the recipient from the channel. Use to un-stick a paid order that never got submitted; confirm it afterward with confirm_order if the store requires confirmation. [#437d68]

Sync a collection to a sales channel (creates/updates the channel-side category and places all products in it that are already synced there). integration_uuid selects which channel; not all channels support collections (e.g. TikTok Shop), which returns a clear "collections_unsupported" error. [#7f246a]

Pull the latest orders from connected fulfillment providers. Pass store_uuid to sync one store; omit it to sync all of your stores. Use to refresh orders that have not come through yet. [#64e731]

Sync one product to a sales channel (WooCommerce/Shopify/Wix) as a listing. PREREQUISITE: the product must first be associated with the store AND synced to its fulfillment provider — call sync_to_fulfillment(product_uuid, store_uuid) FIRST (it does the store association too). If that prerequisite is missing, this tool now AUTO-HEALS it (associate + fulfillment-sync, then retries once) instead of failing with "product not associated with store" — but the clean, explicit order is sync_to_fulfillment then sync_to_channel, and ship_product does the whole pipeline in one call. Defaults to DRAFT — only push live when the user explicitly asks. [#32962d]

Associate a product with a store AND sync it to that store's fulfillment provider (Printful/Printify). This is the REQUIRED step before sync_to_channel: it both puts the product on the store (a product from create_product is standalone) and creates the manufacturing path the sales-channel listing binds to. Run it after the product has variants. [#58f859]

Revert an approved order back to pending so it can be reviewed / re-approved. Only works if the order has NOT yet been submitted to the fulfillment provider. Use to undo an approve_order that was done too early. [#1d050e]

Restore an archived store. It comes back as CLOSED (or ACTIVE if a fulfillment provider is still connected); if it landed CLOSED, connect a provider and call activate_store to reopen it. [#468f04]

Revoke a member's assignment to a workspace (agency / Enterprise). The account owner cannot be unassigned from the Default workspace. Needs an account-wide key. [#7d3ad6]

Remove a product from ONE sales channel, leaving every other channel and the fulfillment provider untouched. The product stays in ApparelHub; only that channel listing goes away. Use this to delist from a single channel — do NOT hand-roll it against the raw unsync endpoint: that endpoint is product-level and defaults to detaching fulfillment AND cascading to every channel, so getting the parameters slightly wrong unsyncs far more than you asked for. To remove a product from EVERYTHING, use archive_product instead. [#bd6407]

Update a collection's name and/or description. A name change is sent to the platform as the collection title. Editing a synced collection marks it for re-sync. [#1d73cf]

Update a product (name, description, price). For a price change that must propagate to synced channels, prefer cascade_price_change. Optionally set tiktok_listing to enrich the TikTok Shop listing (SEO search terms, product highlights, brand, packaging, TikTok-only title/description, and category_id) — applied when the product is synced to a TikTok channel; ignored by other channels. For channel-defined ATTRIBUTES (Material, Style, Washing Instructions...) use set_listing_attributes instead: it validates against the listing category's real schema and tells you which values the channel refused, which this tool cannot. [#c20857]

Update a store's fulfillment workflow / notification settings. Only the fields you pass are changed. fulfillment_mode: "auto" (auto-pilot: paid -> draft -> auto-confirm -> production), "confirm" (auto-draft, you confirm each order), "review" (held before submission for approval). The hold_* guardrails escalate an otherwise-auto/confirm order to a pre-submission review. Set hold_orders_above_amount / hold_below_margin_pct to null to disable that guardrail. hold_channel_risk_review is ON by default and OVERRIDES fulfillment_mode (including "auto"): an order the sales channel is reviewing is never sent to fulfillment while it may still be voided. [#fe3014]

Rename a workspace or archive/unarchive it (agency / Enterprise). The Default workspace cannot be archived. Needs an account-wide key. [#e8ebf4]

Upload artwork the merchant ALREADY OWNS and turn it into a design_uuid usable by create_product / ship_product. This is the way to build products from a client's own files — a logo, a brand mark, a cleared cover, a photograph — instead of generating something new. If a client says their mark must not be redrawn, use this; never regenerate or approximate a mark to work around a missing file. Three ways to supply the file, pick the cheapest one available: 1. `source_url` — an https URL the server can fetch. One call, no context cost. Best when the asset is already hosted or reachable by link (the link must not require sign-in). 2. no source at all — returns a presigned `upload_url` you PUT the bytes to yourself, then call this tool again with the returned `image_uuid` to finish. No context cost, full resolution, and the right choice whenever you can make an HTTP request (curl, fetch, requests). 3. `image_base64` — inline bytes. Works anywhere, but costs roughly 350k tokens per megabyte of file, so reserve it for small files when neither of the above is possible. Accepts PNG, JPEG, WEBP and SVG. SVG is the BEST input for a logo or mark: it is rendered server-side at print resolution, so it stays crisp at any size. Two things must be true of the SVG first — text converted to outlines, and any linked image embedded — otherwise the upload is refused with instructions rather than silently losing that part of the artwork. For pixel art, or any hard-edge raster mark that must stay crisp, pass upscale="pixel" so a small file is enlarged without being smoothed. [#85542c]

Local QC gate for a design: transparency correctness (alpha, clean corners, white premultiply), resolution, and detected text. Returns a 0-100 score + issues. Needs local Python + Pillow. [#8e74ec]

Read the text in a design with local OCR (tesseract) when available, so the agent can confirm spelling. Advisory: pass expected_text to get a match verdict, otherwise the detected text is returned for visual review. has_text is null when OCR is unavailable, meaning UNKNOWN — not "no text"; read the design image yourself in that case. [#57c0af]

QC gate for a rendered product MOCKUP (verify_design_quality checks the design; this checks the render on the garment). Deterministically catches three defects that have actually shipped: an un-keyed chroma-green background printed onto the product, an empty render, and a render too small to judge. It does NOT decide whether the design is upright, clipped, seam-split, or whether every face is printed: those need looking at the image, and a pixel statistic that guessed would be confidently wrong on exactly those cases. It returns a fixed visual_checklist for you to answer by VIEWING the render, so grading is consistent across callers. Treat a clean result as "no hard defect found", not "the mockup is good" until you have answered the checklist. [#feb72a]