create_image_upload
ChatGPTOne-time link to upload an image FILE without its bytes in a tool call (WRITE). Run the returned curl, or give the user upload_url to open. Then call upload_image(upload_id=..., <target>).
Business & Operations
Overview
Orderli is QR, kiosk and online ordering for restaurants, connected to the restaurant's POS. This connector gives Claude the same access you have in the Orderli dashboard, so you can manage your restaurants by asking. What you can do: • Menus: read a menu, add or update items, categories, option groups and cards in bulk, set when a card is available, and translate a menu into your other languages. • POS: check whether the POS is reachable, see how menu items map to POS items, follow a menu import, and inspect the requests Orderli sent for a specific order or payment when something did not arrive. • Restaurant settings: branding, custom CSS for the guest menu, takeaway and delivery times, delivery zones, QR codes and feature flags. • Reporting: revenue and order statistics, sales per dish or table, payment volume, order exports and the orders in a payout. • Receipt templates: design, preview and publish the tickets your tablet prints. • Help: search Orderli's documentation and get direct links to the right dashboard page. Sign in with your Orderli dashboard account. Claude only sees the accounts and restaurants that account can access, and every change goes through the same permissions as the dashboard. Changes are made only when you ask for them; reporting tools are read-only. Requires an Orderli account. Orderli is used by restaurants, cafes, food courts and event caterers in the Netherlands and Europe.
Tools
create_receipt_template
ChatGPTCreate a receipt template seeded with the safe default layout (WRITE). Edit the HTML/CSS with edit_receipt_template, then publish_receipt_template to make it live.
describe_receipt_template_language
ChatGPTDescribe the receipt-template language: engine, docs link, and variables. Returns which engine renders the ticket (Jinja2, sandboxed), a link to its syntax docs, the rules that apply, and the list of variables a template can read — each with a dotted path, type and example. The variable list comes from the real render context, so it stays current as the data model changes. Read this before writing or editing a template's HTML.
edit_receipt_template
ChatGPTPatch a template's HTML or CSS with optimistic concurrency (WRITE). Read first with get_receipt_template; pass its digest as expected_digest. The HTML is Jinja2 (sandboxed) — call describe_receipt_template_language for the syntax and the variables it can read. target is "html" or "css". Edits are all-or-nothing find/replace or append rows. Saves a new draft version; publish when ready. dry_run previews only.
edit_restaurant_css
ChatGPTPatch a restaurant's Custom CSS with optimistic concurrency (WRITE). Read first with get_restaurant_css and pass its expected_digest. Edits are all-or-nothing find/replace or append rows, e.g. {"find": "...", "replace": "..."} or {"append": "...css..."}. Use dry_run to preview. After styling changes, visually check the live guest menu.
export_orders
ChatGPTExport every order group in a date range (auto-paginated). Pulls all pages of the orders list for the range and returns a single flat `
orders array, ready to render as CSV/spreadsheet. Each row is a slim summary by default; pass detail=True for the full order groups. A range too big to return at once comes back with next_offset; pass it as offset to collect the rest. Args: restaurant_id: the restaurant to export. start_date: start date, YYYY-MM-DD (inclusive). end_date: end date, YYYY-MM-DD` (inclusive). max_records: hard cap on rows returned (default 2000). payment_status: optional status filter. payment_method: optional method filter. detail: return full order groups instead of slim summaries. offset: resume the export at this row; see next_offset in the reply. account_id: optional; resolved from restaurant_id if omitted.find_accounts
ChatGPTFind accounts the current token can access (one slim row each). Each row is id, name, plan, role and a restaurant count. Use `
find_restaurants to list an account's restaurants, and get_restaurant_config` for the full configuration. Args: query: case-insensitive substring to match against the account name (or an exact account id). Omit to list from the top. limit: max rows to return (default 25). offset: skip this many matches — walk the full list a page at a time.find_restaurants
ChatGPTFind restaurants the token can access (one slim row each). Restaurants are nested in the accounts payload. Each row is id, name and owning account — enough to match a name to an id. Call `
get_restaurant_config` for the full configuration and a deep link. Args: account_id: optional account id to filter to. query: case-insensitive substring to match against the restaurant name (or an exact restaurant id). Omit to list from the top. limit: max rows to return (default 25). offset: skip this many matches — walk the full list a page at a time.get_dashboard_link
ChatGPTBuild a link to a restaurant's dashboard page — no data is fetched. Use this to hand the user a clickable destination when they want to see something in the dashboard, even if a data call wasn't approved or isn't needed. The returned `
url is ready to present as a markdown link. Marketplaces (food courts): to show every stand at once on the analytics page — e.g. "revenue or service fee per stand" — pass the marketplace parent's restaurant_id with page="analytics", locations=["all"] and per_location=True. Args: restaurant_id: the restaurant (or marketplace parent) to link to. page: which page — one of analytics, orders, payments, menu, pos-items, settings, pos, payment-settings, api-tokens, devices, activity-log, account. Defaults to analytics. account_id: optional; resolved from restaurant_id when the page needs it. locations: analytics only — restaurant ids to pre-select in the location selector, or ["all"] for every location in the account. per_location: analytics only — open the per-location breakdown. start: optional YYYY-MM-DD start date hint for the date range. end: optional YYYY-MM-DD` end date hint for the date range.get_delivery_zones
ChatGPTList delivery zones. Money fields are backend cents.
get_doc
ChatGPTFetch the full text of a doc chunk or operation by its doc_id. Args: doc_id: an id from search_docs, e.g. "md:CLAUDE.md#3" or "op:get_...".
get_endpoint
ChatGPTGet the full OpenAPI operation (params, request/response schema, errors). Args: operation_id: e.g. "get_orders.get_orders" from search_endpoints.
get_import_status
ChatGPTGet the POS import pipeline status (state, percent complete, timestamps). Args: restaurant_id: the restaurant whose import status to fetch. account_id: optional; resolved from restaurant_id if omitted.
get_menu
ChatGPTGet a restaurant's Orderli menus, cards, categories, option groups and items. Menus are returned in full: every per-menu setting (text_alert, service_fee, takeaway, pickup_codes, custom_class, menu_times, ...) lives there. Other rows are projected to identifying fields; a row is active unless it says active:false. Only the first page carries menus, cards, categories and option groups; later pages and queries carry items. Args: restaurant_id: the restaurant whose menu to fetch. account_id: optional; resolved from restaurant_id if omitted. menu_id: scope to one menu and the cards/items it reaches. query: a term or list of terms; items whose name contains one or whose id equals one. unmatched lists terms that found nothing. ids: return these menus/cards/categories/items/option groups in full (translations, times, prices, linked ids) instead. detail: raw rows instead of projections. limit: max items per page (default 500). offset: skip this many items; see next_offset in the reply.
get_menu_changes
ChatGPTGet the menu change history for a restaurant (newest first, paginated). Returns `
{page, page_results, total_results, data: [...]}`. Args: restaurant_id: the restaurant whose change log to fetch. account_id: optional; resolved from restaurant_id if omitted. page: 1-based page number. results: page size (default 50).get_order_pos_requests
ChatGPTPOS requests Orderli made while pushing one order to the POS. Returns `
{requests: [...]}, newest first. Each entry has the service (e.g. mpluskassa, untill, lightspeed), the endpoint hit, the request/response bodies, and the requested_at/responded_at timestamps. The first place to look when an order didn't reach the POS. Find the order_group_id via list_orders. Request/response bodies are previewed (truncated, with request_truncated/request_length set) to keep the log small; pass request_id (an id` from the list) to get that one entry with its full, untruncated bodies. Args: restaurant_id: the restaurant the order belongs to. order_group_id: the order group whose POS requests to fetch. request_id: return only this entry, with full request/response bodies. account_id: optional; resolved from restaurant_id if omitted. limit: max entries per page (default 50). offset: skip this many entries; see next_offset in the reply.get_orders_timeseries
ChatGPTOrder counts over time, bucketed by timeframe — the orders chart series. Args: restaurant_id: the restaurant to report on. start: start date, `
YYYY-MM-DD (inclusive). end: end date, YYYY-MM-DD (inclusive). timeframe: bucket size — hour, day (default), week, month`. account_id: optional; resolved from restaurant_id if omitted. limit: max buckets per page (default 500). offset: skip this many buckets; see next_offset in the reply.get_payment_pos_requests
ChatGPTRequests Orderli made to the POS/PSP while processing one "Payment". "Payments" are the pay-later product, which is being phased out — they are NOT the general order/transaction list. For an order's POS requests use `
get_order_pos_requests with an order_group_id from list_orders; only use this tool for a legacy pay-later payment_id (taken from the dashboard's Payments page). Returns {requests: [...]}, newest first, with the same shape as get_order_pos_requests: service (e.g. ccv, mollie, stripe, mpluskassa), endpoint, request/response, and timestamps. Bodies are previewed; pass request_id` to get one entry with its full, untruncated bodies. Args: restaurant_id: the restaurant the payment belongs to. payment_id: the pay-later payment whose requests to fetch. request_id: return only this entry, with full request/response bodies. account_id: optional; resolved from restaurant_id if omitted. limit: max entries per page (default 50). offset: skip this many entries; see next_offset in the reply.get_payments_breakdown
ChatGPTPayment volume over time: money collected through Orderli. `
results combines checkout orders paid online and pay-later bill payments (both including tips); payment_sources splits them into paid_order_groups and pay_later_payments. Cash, invoice, external-terminal and test orders are excluded, so this is NOT total order revenue. For that use get_revenue_stats / get_revenue_breakdown / get_orders_timeseries. Args: restaurant_id: the restaurant to report on. start: start date, YYYY-MM-DD (inclusive). end: end date, YYYY-MM-DD (inclusive). timeframe: bucket size — hour, day (default), week, month`. account_id: optional; resolved from restaurant_id if omitted. limit: max buckets per page (default 500). offset: skip this many buckets; see next_offset in the reply.get_pos_mapping
ChatGPTGet the resolved 'POS order details' mapping for the active POS. Args: restaurant_id: the restaurant whose mapping to fetch. account_id: optional; resolved from restaurant_id if omitted.
get_pos_status
ChatGPTPOS reachability for a restaurant (last reachable time + last error). Args: restaurant_id: the restaurant whose POS status to check. account_id: optional; resolved from restaurant_id if omitted.
get_receipt_template
ChatGPTRead one receipt template's current config (HTML, CSS) with a digest. The `
digest` is what edit_receipt_template requires as expected_digest — it names the version an edit was written against. digest_only returns just the digest and field lengths — a cheap way to ask "has it changed?" without pulling the markup again.get_restaurant_config
ChatGPTGet the full configuration for a restaurant. Includes POS integration fields, payment settings, opening times, service config, locale/currency, feature flags, and more. Args: restaurant_id: the restaurant to inspect. account_id: optional; resolved from restaurant_id if omitted.
get_restaurant_css
ChatGPTRead a restaurant's Custom CSS, and nothing else. Returns the stylesheet with a `
digest of its contents, which edit_restaurant_css` requires — it is how an edit says which version it was written against. Args: restaurant_id: the restaurant whose stylesheet to read. account_id: optional; resolved from restaurant_id if omitted. digest_only: return just the digest and length — a cheap way to ask "has it changed?" without pulling the sheet again.get_revenue_breakdown
ChatGPTRevenue and sales volume per dish, option, menu, table or area over a date range (YYYY-MM-DD, inclusive), biggest revenue first. For best sellers / items sold / quantity per dish use per_dish: quantity_sold is units, not orders. per_option_chosen lists paid options only. Menu, table and area rows count orders. Rows: {id?, label, revenue, quantity_sold|orders}; page with offset (next_offset).
get_revenue_stats
ChatGPTRevenue summary cards for a date range (total revenue, order count, tips, average order value, …) with the previous-period comparison. Args: restaurant_id: the restaurant to report on. start: start date, `
YYYY-MM-DD (inclusive). end: end date, YYYY-MM-DD` (inclusive). account_id: optional; resolved from restaurant_id if omitted.get_settlement_orders
ChatGPTThe order groups settled in a specific payout/settlement (paginated). Args: restaurant_id: the restaurant the payout belongs to. payout_id: the settlement/payout id (see get_restaurant reports). page: 1-based page number. page_size: items per page (default 100, max 1000). account_id: optional; resolved from restaurant_id if omitted.
inspect_guest_menu_entrypoint
ChatGPTOpen a real Orderli guest URL, call guest /init + /menu, and return compact visible IDs. Use a URL returned by setup_qr or list_guest_entrypoints when available. Restaurant ids and menu ids are not guest URL path segments. In auto/readiness mode, ordering sessions return required next steps. A 404, guest_error, or api_error means guest readiness is blocked.
list_api_tokens
ChatGPTList the API tokens configured for a restaurant (names + permissions). Args: restaurant_id: the restaurant whose tokens to list. account_id: optional; resolved from restaurant_id if omitted.
list_guest_entrypoints
ChatGPTReturn candidate public guest URLs; discovery only, not verification. Reads restaurant database_name and QR codes so agents do not invent URL shapes. For readiness, inspect a returned URL, probe checkout if relevant, then browser-test.
list_orders
ChatGPTOne page of order groups for a restaurant (newest first). Returns `
{page, page_results, total_results, data: [...]}. Each row is a slim summary by default; pass detail=True for the full order group. Use the filters to narrow, or export_orders to pull a whole date range at once. Args: restaurant_id: the restaurant whose orders to list. start_date: filter from this date, YYYY-MM-DD (optional). end_date: filter to this date, YYYY-MM-DD (optional). page: 1-based page number. results: page size (default 100, max 100). payment_status: filter by status (e.g. success, open, failed). payment_method: filter by method (e.g. ideal, card, cash). order_type: filter by type (normal, delivery, takeaway, takeaway_onsite, kiosk`). has_tip: only orders that include a tip. has_coupon: only orders that used a coupon. detail: return the full order group instead of the slim summary. account_id: optional; resolved from restaurant_id if omitted.list_pos_import_requests
ChatGPTPOS requests Orderli made while importing the menu/catalog from the POS. Returns `
{requests: [...]}, newest first, each with service, endpoint, request/response and timestamps. Reach for this when a menu import failed or pulled in the wrong items. Unlike the order/payment request logs this is restaurant-wide — no order or payment id needed. Import payloads are large, so bodies are previewed; pass request_id` to get one entry with its full, untruncated bodies. Args: restaurant_id: the restaurant whose import requests to fetch. request_id: return only this entry, with full request/response bodies. account_id: optional; resolved from restaurant_id if omitted. limit: max entries per page (default 50). offset: skip this many entries; see next_offset in the reply.list_pos_items
ChatGPTList the POS's own items, the layer menu items map onto. pos_id is the POS's id: a menu item's pos_id must equal it. pos_item_id is Orderli's row id for the POS item, never a POS id. Args: restaurant_id: the restaurant whose POS items to list. account_id: optional; resolved from restaurant_id if omitted. query: a term or list of terms (resolve many dishes in one call); each matches a name substring or an exact pos_id/pos_item_id. unmatched lists terms that found nothing. options: add each row's option groups and their options. limit: max rows per page (default 100). offset: skip this many rows; see next_offset in the reply.
list_pos_payment_methods
ChatGPTList the payment methods the POS exposes for a restaurant. Args: restaurant_id: the restaurant whose POS payment methods to list. account_id: optional; resolved from restaurant_id if omitted.
list_receipt_render_logs
ChatGPTRead recent receipt render attempts for debugging a misprint. Each row records whether a tablet render of an order succeeded, which template/version it used, and any templating error. Filter by template_id, order_group_id, or only_failures to find what went wrong after the fact.
list_receipt_templates
ChatGPTList a restaurant's receipt templates (name, ticket type, publish state).
preview_receipt_template
ChatGPTRender a template against a sample or a real order and return the HTML. Use this to check how order data flows through the layout before publishing. Pass order_group_id to render a real order from this restaurant; otherwise sample_key picks a built-in sample (the response lists the available ones; omit both for the default multi-group sample). Search real orders with search_orders_for_receipt_preview.
publish_receipt_template
ChatGPTPublish a template version so tablets start printing it (WRITE). Defaults to the latest draft version. Publishing runs a render gate against sample orders; a template that cannot render is refused rather than made live, so printing never breaks.
resolve_restaurant
ChatGPTResolve a restaurant_id to its owning account_id (plus basics). The Orderli dashboard API is account-first: almost every path is nested under `
/accounts/{account_id}/..., so a bare restaurant id is not enough to build one. There is no /restaurants/{id} collection to guess at. Call this first when all you have is a restaurant id — it hands back the account_id` every other tool needs. Returns id, name, account_id, account_name and pos_integration, plus a dashboard deep link. Prefer this over guessing restaurant-first paths. Args: restaurant_id: the restaurant to resolve.scan_qr
ChatGPTRead every QR code in an image and return the data each one encodes (READ). A standalone reader with no knowledge of Orderli: give it the URL of an image — e.g. an uploaded photo of a QR sticker — and it returns the exact text/URL each QR code contains. It does not act on the result; you decide what to do with the values. For an Orderli table sticker that means passing the decoded URL to setup_qr. A photo containing several stickers returns several codes. Decoding is expensive, so this is rate limited.
search_docs
ChatGPTSearch Orderli docs and API operations by keyword. Args: query: natural-language or keyword query. max_results: max hits to return. source: optionally restrict to "markdown" or "openapi".
search_endpoints
ChatGPTSearch available API operations by keyword (path, summary, tags). Returns matching operationIds with method + path. Use get_endpoint for the full request/response schema.
search_orders_for_receipt_preview
ChatGPTFind real orders to preview a template against (id, table, pickup, name). Matches order id, table number, pickup code or guest name; empty query returns the most recent orders. Feed an id to preview_receipt_template's order_group_id.
set_card_schedule
ChatGPTSet when a card (section group) is visible and orderable (WRITE). A card has two independent schedules: - visibility_windows: when guests can SEE the card. - ordering_windows: when guests can ORDER from it. Omit to match visibility. Make ordering NARROWER than visibility for "view-only until X" — e.g. a breakfast card visible from 08:00 but only orderable 08:00-11:00, or a whole menu you can browse before it opens. Each window is {"days": "all", "weekdays", "weekends", "mon-fri", a day name/list of names, or 0-6 where Mon=0; "start": "HH:MM", "end": "HH:MM"}. Prefer day names/ranges over numeric days.
set_delivery_availability
ChatGPTOpen or close guest delivery for whole day(s) (WRITE). Delivery capacity is driver-backed, so this creates or removes a 'close' capacity override on the restaurant's delivery driver resource. Use it for requests like "my driver called in sick, close delivery tomorrow" rather than editing menus, zones or ordering settings.
available is required: False closes delivery, True reopens it (removes the matching close override). Give the day as either date (YYYY-MM-DD) or days_from_today (0 = today, 1 = tomorrow) so you never have to know the calendar date; end_date (YYYY-MM-DD, inclusive) spans a multi-day range. Days are resolved in the restaurant's own timezone. Takeaway and dine-in are unaffected.set_menu_links
ChatGPTSet the ordered children of one or many menu containers (WRITE). links maps each parent id to its child ids in display order, e.g. {"178143": [11248, 11257], "178145": []}. Each list replaces that parent's whole set ([] clears it); parents left out are untouched. Children must already exist (upsert_* tools). parent_type is one of "menu.cards", "card.categories", "category.items", "item.option_groups" (option-group ids), "option_group.items" (item ids).
set_qr_kiosk_dine_choice
ChatGPTSet or clear a kiosk QR's fixed eat-in/take-away choice (WRITE).
set_restaurant_feature_flag
ChatGPTEnable or disable a feature flag for a restaurant (WRITE). Upserts the restaurant↔flag link (creating it if needed, un-deleting it if it was removed). Requires ORDERLI_ALLOW_WRITES and a write-scoped token. Args: restaurant_id: the restaurant to toggle the flag on. flag: the flag slug (e.g. "rootline_checkout_components") or its numeric id. enabled: True to enable (default), False to disable. account_id: optional; resolved from restaurant_id if omitted.
setup_qr
ChatGPTRegister an existing printed QR sticker for table/takeaway/delivery/kiosk (WRITE). Needs the sticker's code: pass qr_link (the URL it encodes) or code; it cannot mint one. For takeaway/delivery ordering without a sticker use the URLs from list_guest_entrypoints. area can create/use an area and colour the QR. Returns candidate guest URL(s), not readiness evidence.
translate_menu
ChatGPTFill in a menu's other languages with Orderli's AI translator (WRITE). Orderli has a menu-tuned translation SPECIALIST behind this — strongly prefer it over hand-translating rows. Build the whole menu in the single language you actually have (source_language), then call this once to add every target language across all items, categories and cards. This is the right way to get a bilingual menu — do NOT paste one language's text into another language's fields. Runs as an async job and may take a minute or two on a large menu; this tool waits for it and applies the result. If it is still running when the wait budget is up, it returns a job_id — call translate_menu again with that job_id to resume waiting and apply. only_add_missing=True (default) only fills empty translations and never overwrites text you wrote. entity_types defaults to ["item","category","card"]. account_id resolves from restaurant_id when omitted.
update_legacy_menu_custom_css
ChatGPTUpdate CSS that already exists on a legacy menu (WRITE, deprecated). New menu-level CSS is never allowed. Prefer setting `
custom_class on the menu and adding scoped rules to the restaurant-level Custom CSS via update_restaurant_settings. When a menu already has legacy CSS, the first non-empty update returns a deprecation warning; retry with force=True only after the user explicitly approves the emergency override. Clearing existing CSS is allowed without force. Pass exactly one of: - custom_css: the complete replacement, or "" to clear it. - append_css`: CSS to append while preserving the existing value. Args: restaurant_id: restaurant that owns the menu. menu_id: existing menu whose legacy CSS is being changed. custom_css: complete replacement, or an empty string to clear. append_css: CSS to append to the existing value. force: confirm an urgent update after reviewing the warning. account_id: optional; resolved from restaurant_id if omitted.update_ordering_settings
ChatGPTUpdate takeaway/delivery switches, lead times and timeslots (WRITE). Lead-time and timeslot fields are minutes. Restaurant-level fallback delivery_costs/minimums are EUR; zone-specific fees go through upsert_delivery_zone.
update_restaurant_settings
ChatGPTUpdate restaurant branding/look-and-feel fields (WRITE). Partial update: only passed fields change. Use this for display name, image ids/URLs, colors, review URLs/questions, or full custom_css replacement. For takeaway/delivery switches and lead times use update_ordering_settings instead.
upload_image
ChatGPTUpload an image as the COMPLETE image workflow (WRITE). Do not upload first and then update the entity separately. Pass one source and at most one target: menu_item_id=266689 (product photo), category_id, card_id (+ card_image_field="header_image"), or restaurant_id + set_as="logo"| "hero_image"|"article_placeholder_image"|"background_image" (guest menu background; menu_id picks the menu). Source: image_url, upload_id from create_image_upload for a file, or image_id of an image already uploaded. Success includes image_id; apply_error means upload worked but attach failed.
upsert_delivery_zone
ChatGPTCreate/update delivery zones (WRITE). Postal_codes are prefixes. Backend units are cents/seconds, but delivery_fee_eur, min_*_eur and duration_minutes are accepted. Pass zones=[...] for bulk; omit zone_id to create, pass it to update.
upsert_menu
ChatGPTCreate/update a top-level menu (WRITE). Menu tree: menu -> cards -> categories -> items. Use card_ids or set_menu_links("menu.cards") to attach cards. menu_times defaults to always-on; rows use day plus start/end "HH:MM", one row per day. Online: takeaway and/or delivery (thuis derived). Only if the user asked: change_menu_type / allow_multiple_types. Pass menus=[...] for bulk.
upsert_menu_card
ChatGPTCreate/update menu cards, singly or in bulk (WRITE). Bulk rows use cards=[{...}]. Omit card_id to create; pass it to update only supplied fields. A card holds categories and sits on menus (set_menu_links "card.categories" / "menu.cards"). It may be nameless; name can generate translations rows {"language","name","description"?}. Colors are hex strings.
upsert_menu_category
ChatGPTCreate/update menu categories, singly or in bulk (WRITE). Bulk rows use categories=[{...}]. Omit category_id to create; pass it to update only supplied fields. Guest text belongs in translations; name can generate default translations. Normal orderable sections default to style="single"; display-only styles include message, banner, qr-link, reservation and image-only. Link items/cards with set_menu_links. style="qr-link" switches the guest to target_qr_id's menu (target_qr_table_source "linked" or "scanned"). Visibility: category_times [{day: -1 daily or 0-6 Mon=0, start, end}] in seconds of the day, or visibility_schedule_set_id on unified scheduling.
upsert_menu_item
ChatGPTCreate/update menu items, singly or in bulk (WRITE). Bulk rows use items=[{...}]. Omit item_id to create; pass it to update only supplied fields. Guest text goes in translations (name can seed them). price is VAT-inclusive, default 0; vat is a fraction like 0.09; open_price=true: staff type the price per order. Write the source language, then translate_menu for the others. Link categories, modifiers and images with set_menu_links/upload_image.
upsert_option_group
ChatGPTCreate/update option groups (modifiers), singly or in bulk (WRITE). Bulk rows use option_groups=[{...}]. Omit option_group_id to create; pass it to update only supplied fields. type is "select" (pick one) or "addons" (pick several). Labels are translations rows {"language","title"} ("name" maps to title); name can generate them. Fill with set_menu_links("option_group.items"), attach to dishes with set_menu_links("item.option_groups").
whoami
ChatGPTReport the server's configuration and the reach of the current token. Returns how much this token can see, not the inventory itself: an admin token reaches the whole customer base, and listing it costs tens of thousands of tokens to answer a question nobody asked. Use `
find_accounts / find_restaurants` to look up ids.create_image_upload
Claudecreate_receipt_template
Claudedescribe_receipt_template_language
Claudeedit_receipt_template
Claudeedit_restaurant_css
Claudeexport_orders
Claudefind_accounts
Claudefind_restaurants
Claudeget_dashboard_link
Claudeget_delivery_zones
Claudeget_doc
Claudeget_endpoint
Claudeget_import_status
Claudeget_menu
Claudeget_menu_changes
Claudeget_order_pos_requests
Claudeget_orders_timeseries
Claudeget_payment_pos_requests
Claudeget_payments_breakdown
Claudeget_pos_mapping
Claudeget_pos_status
Claudeget_receipt_template
Claudeget_restaurant_config
Claudeget_restaurant_css
Claudeget_revenue_breakdown
Claudeget_revenue_stats
Claudeget_settlement_orders
Claudeinspect_guest_menu_entrypoint
Claudelist_api_tokens
Claudelist_guest_entrypoints
Claudelist_orders
Claudelist_pos_import_requests
Claudelist_pos_items
Claudelist_pos_payment_methods
Claudelist_receipt_render_logs
Claudelist_receipt_templates
Claudepreview_receipt_template
Claudepublish_receipt_template
Clauderesolve_restaurant
Claudescan_qr
Claudesearch_docs
Claudesearch_endpoints
Claudesearch_orders_for_receipt_preview
Claudeset_card_schedule
Claudeset_delivery_availability
Claudeset_menu_links
Claudeset_qr_kiosk_dine_choice
Claudeset_restaurant_feature_flag
Claudesetup_qr
Claudetranslate_menu
Claudeupdate_legacy_menu_custom_css
Claudeupdate_ordering_settings
Claudeupdate_restaurant_settings
Claudeupload_image
Claudeupsert_delivery_zone
Claudeupsert_menu
Claudeupsert_menu_card
Claudeupsert_menu_category
Claudeupsert_menu_item
Claudeupsert_option_group
Claudewhoami
ClaudeCapabilities
Writes
Example Prompts
Click any prompt to copy it.
Links
App Stats
122
Tools
3
Prompts
ChatGPT, Claude
Platforms
Category
Grocery ShoppingWorks with
ChatGPT
Claude
Data refreshed daily