MCP App Store
by KeenEthics
Jinko icon

Jinko MCP Server for ChatGPT

By Jinko SAS

Overview

Get your personal travel assistant to search and book flights, hotels, and car rentals for you. Tell Jinko where you want to go, your budget, or your travel preferences. Jinko finds the best options and helps you book everything in one conversation. What Jinko can do for you: 1. Predict flight prices and tell you the best time to book. 2. Find the cheapest dates to travel and save money. 3. Discover and compare destinations based on your budget and preferences. 4. Monitor flight prices and help you rebook when prices drop. From search to booking, Jinko handles it all: 1. Book flights, hotels, and car rentals securely with Stripe Link. 2. Change or cancel your bookings whenever you need to.

How to connect

  1. ChatGPT: Open the Jinko listing in the ChatGPT app directory and choose Connect. ChatGPT listing Jinko exposes 30 tools here.

Tools

car_cancel_commit

ChatGPT
Cancels a booked car rental using the "ccl_…" quote_id from car_cancel_preview. Call only after the customer has seen the preview and explicitly approved cancellation. Guest verification: booking_ref is the Jinko JNK-XXXXXX reference (not an airline record locator or a hotel confirmation number) and last_name is the primary traveler's last name; no login is required. Call get_booking first, choose the items[] entry to service and pass its booked_item_id. Use the same booking_ref + booked_item_id + last_name on every step; booked_item_id stays the same after an exchange. For a manual_required preview, send manual_ok: true only after the customer explicitly agrees to a Jinko agent completing it. Never infer that consent or invent an amount. A changed or expired preview is refused without cancelling and must be shown again. Next step: call car_cancel_status with the same booking identity and booked_item_id. Cost: 1 credit per call.

car_cancel_preview

ChatGPT
Prices cancellation of a booked car rental. It changes nothing. Guest verification: booking_ref is the Jinko JNK-XXXXXX reference (not an airline record locator or a hotel confirmation number) and last_name is the primary traveler's last name; no login is required. Call get_booking first, choose the items[] entry to service and pass its booked_item_id. Use the same booking_ref + booked_item_id + last_name on every step; booked_item_id stays the same after an exchange. Returns the fee, refund, and a "ccl_…" quote_id. fee_known: false means the amount remains to be confirmed, not that cancellation is free. A manual_required result needs explicit customer consent before a Jinko agent may complete it. Next step: show the customer the result, obtain explicit approval, then call car_cancel_commit with the quote_id. Cost: 1 credit per call.

car_cancel_status

ChatGPT
Reads the latest car-rental cancellation attempt. It changes nothing and takes no operation handle. Guest verification: booking_ref is the Jinko JNK-XXXXXX reference (not an airline record locator or a hotel confirmation number) and last_name is the primary traveler's last name; no login is required. Call get_booking first, choose the items[] entry to service and pass its booked_item_id. Use the same booking_ref + booked_item_id + last_name on every step; booked_item_id stays the same after an exchange. Use the same booking_ref, last_name, and booked_item_id used for preview. A pending result is still being completed; poll this tool rather than committing again. Cost: 1 credit per call.

checkout

ChatGPT
Checkout a trip — returns the web checkout_url a human opens to pay and, when available, agent_spt_params an agent can use to pay programmatically (no browser). PREREQUISITES (call these first via the trip tool): trip(add_item) → add the chosen flight or hotel to the trip trip(upsert_travelers) → set traveler details + contact info USAGE: checkout({ trip_id }) Jinko prices the trip (schedules a quote and waits until it completes) in one synchronous call. Checkout books nothing and charges nothing: the booking starts only when the traveller pays at checkout_url or you call submit_agent_payment. • Human path: open the returned checkout_url in a browser to pay. Stripe webhooks finalize the booking after payment succeeds; no client-side confirm step is required. • Agent path, only when agent_spt_params is present: mint a Shared Payment Token scoped to those params, then call submit_agent_payment({ trip_id, shared_payment_token }). If agent_spt_params is absent, use checkout_url. OPTIONAL FLIGHT ANCILLARIES: checkout → trip(select_ancillaries) → checkout. The first checkout quotes the trip; use get_trip to inspect available_ancillaries, select with the user's approval, then checkout again to refresh the quote and payment artifacts. Pay only from the latest checkout response. If travelers are edited after checkout, the quote is invalid: call checkout again before paying. IMPORTANT: Never fabricate traveler data. Always make sure trip(add_item) and trip(upsert_travelers) have been called with real user-provided data first. CHILDREN: checkout refuses a trip whose flight is priced for a child or an infant (error_code TRAVELERS_ON_JINKO_PAGE) and returns trip_link: that trip's travelers are entered, and it is priced and paid, on that Jinko page. Give the user the link; do not collect traveler details for that trip. Cost: 1 credit per call.

claim_best_price_guarantee

ChatGPT
Best Price Guarantee: makes a claim — asks for a Jinko hotel offer at a lower price the traveller found on another site. Nothing is booked or charged. WHEN TO USE: - The traveller compared a Jinko hotel offer with the same stay on another site and found it cheaper there. - The traveller has seen a hotel elsewhere and has no Jinko offer yet: run hotel_search first (it can search by hotel name), let them pick the offer, then claim with its trip_item_token. HOW TO CALL: - trip_item_token (the htl_… offer from hotel_search), reference_url, reference_screenshot_url, reference_total, reference_currency. - wait_seconds (0–20, default 15) is how long the call waits for the result; most claims are answered within it. A "pending" answer carries the claim's id: read its result with get_best_price_guarantee. WHAT THE TRAVELLER MUST SUPPLY: - The link to the page showing the lower price — the hotel's own page on that site, for the same dates and guests — and, on most accounts, an https address a screenshot of that page can be downloaded from. Both come from the traveller. Never build, guess or reuse one. Jinko opens the link and checks the hotel, the dates, the guests, the room, the board, the cancellation terms and the price. - The total they would pay at booking on that page, in the currency of the Jinko offer. Taxes payable at the property are not part of it. A price in another currency is declined, not converted. WHAT COMES BACK: - "offered": a new offer (offer.trip_item_token, starting with htl_neg_) at offer.price_money, valid until offer.expires_at. offer.saving_money is the saving against the rate's current standard price (offer.original.total_money, which can be lower than the price the earlier search showed); 0 means that current price was already at or below the reference. - "declined": no offer, with decline.reason, and decline.detail when the check of the reference page found something to explain — pass it on to the traveller. The offer from hotel_search can usually still be booked at its own price. - "pending": the review is still running; reading the reference page can take a few minutes. It is not a decline and not a failure — read the result with get_best_price_guarantee; do not tell the traveller it failed. - "expired": the offer was not used in time. BOOKING THE NEW OFFER: - Call trip with add_item.trip_item_token = offer.trip_item_token, then checkout, exactly as for any hotel offer. - A trip that holds the new offer can hold nothing else. Start a new trip (no trip_id); or, when the trip holds only the offer being replaced, send remove_item for that item and add_item for the new offer in the same trip call. Adding it beside another item is refused. - The price at checkout can move with the supplier. WORDING: this is the price Jinko can offer after its own review. Do not tell the traveller that Jinko negotiated with the hotel or contacted it. A claim can be declined, so do not promise a match before the result says "offered". Cost: 0 credits per call.

find_dates

ChatGPT
Find the best date options for a known origin and destination using cached pricing. Takes the same inputs as flight_calendar except offset, but returns one cheapest itinerary per (departure, return) date-pair, spread across the requested departure window (20 date-pairs by default; limit adjusts), so the user sees a diverse set of "when should I fly?" options rather than the full per-date list. WHEN TO USE THIS TOOL: - The user knows the route (origin AND destination) and is flexible on WHEN, and wants a short, spread-out shortlist of the best date options. - Examples: "When is the cheapest time to fly Paris → New York in June?", "Best weekends to go from London to Lisbon next month", "Give me a few good date options for a week in Tokyo". WHEN TO USE flight_calendar INSTEAD: - The user wants the full price-per-date list / calendar grid for a route, not a curated shortlist. WHEN TO USE flight_search INSTEAD: - The user wants live pricing on one specific route with an exact departure date. For a round trip, also require an exact return date; one-way trips need no return date. - Examples: "Paris → NYC one-way on June 17", "Paris → NYC, June 17 → June 26" - flight_search hits live pricing (each call has a cost) and is the step immediately before booking. Use once the route and required travel dates are locked. IMPORTANT: All dates MUST be in the future. Never use past dates. Origin and destination: the city code when the user named a city (PAR, NYC), the airport code when they named an airport (CDG, JFK). A code that names a multi-airport city covers every airport of that city; any other code is that one airport. LAX, MIA and LAS are both a city code and that city's main airport, so they are read as the city here. By default search roundtrip unless the user explicitly asks for one-way (trip_type="oneway"). Flow: find_dates → flight_search (price-check the chosen dates) → trip → checkout The result carries search_criteria (what was actually searched) and next (the call to make next — on an empty answer up to three exact-date flight_search searches, on a non-empty one price_check for the best candidates). Follow next before telling the user there are no flights or quoting a cached price. Cost: 1 credit per call.

find_destination

ChatGPT
Discover travel destinations accessible from your departure airports. Search globally or restrict results with the destinations list of IATA airport/city codes. For a region or theme, supply the relevant destination codes; there is no region parameter. Returns destinations sorted by cheapest available flight. ORIGINS: Use 3-letter IATA airport codes (CDG, JFK) or city codes (PAR, NYC). When listing airports, include the nearby airports the user can use (e.g., CDG + ORY for Paris). A code that names a multi-airport city covers every airport of that city; any other code is that one airport. LAX, MIA and LAS are both a city code and that city's main airport, so they are read as the city here. Next step: Use flight_calendar with the destination to get specific flights with offer_tokens for pricing. The result carries search_criteria (what was actually searched) and, on a non-empty answer, next: the flight_calendar calls for the best destinations. Follow next to price a destination; on an empty answer the message names the window that was searched. Every price this tool returns is one adult's cached fare (search_criteria.priced_for: "one_adult"); the tool takes no party. To price the whole party, price-check the offer using flight_search with price_check and the travellers' counts in traveler_counts. Cost: 1 credit per call.

flight_calendar

ChatGPT
Search flights between a known origin and destination using cached pricing, especially for flexible dates. WHEN TO USE THIS TOOL (CRITICAL): - The user provides both an origin AND a destination (city or airport) - Examples: "Paris to Barcelona", "JFK to CDG", "London to NYC for a weekend" - Supports loose / flexible dates: single dates, date arrays, date ranges, stay_days - ALSO the right tool for "cheapest flight", "best flight", "find me a flight", "cheapest date" phrasings — returns the cheapest cached itineraries. WHEN TO USE find_destination INSTEAD: - The user does NOT specify a destination: "Where should I go from Paris?", "Best deals from NYC" - The user wants inspiration: "Beach destinations from London", "Cheap flights from SF" WHEN TO USE find_dates INSTEAD: - The user wants a short shortlist of the best date options rather than the full per-date list: one cheapest itinerary per (departure, return) date-pair, spread across the departure window. Same inputs as this tool except offset (find_dates does not paginate; limit scales the shortlist). WHEN TO USE flight_search INSTEAD: - The user wants live pricing on one specific route with an exact departure date. For a round trip, also require an exact return date; one-way trips need no return date. - Examples: "Paris → NYC one-way on June 17", "Paris → NYC, June 17 → June 26" - flight_search hits live pricing (each call has a cost) and is the step immediately before booking. Use once the route and required travel dates are locked. IMPORTANT: All dates in query parameters (departure_dates, departure_date_ranges, return_dates, return_date_ranges) MUST be in the future. Never use past dates. Use departure_dates for specific days or departure_date_ranges for a window. There is no departure_date parameter. Origin and destination: send the city code when the user named a city (Paris → PAR, New York → NYC) and the airport code when they named an airport (JFK, CDG). A code that names a multi-airport city covers every airport of that city; any other code is that one airport. LAX, MIA and LAS are both a city code and that city's main airport, so they are read as the city here — this tool has no type field; flight_search has origin_type / destination_type to pin one airport. ROUTE SEARCH: - Use exact 3-letter IATA airport codes or IATA city code for both origin and destination - Date ranges OR stay duration for flexible trip planning - Natural trip duration (stay_days) instead of exact return dates - By default, please search roundtrip flights unless user specifies one-way. Use trip_type="oneway" ONLY when the user explicitly asks for a one-way trip USE CASES: ✓ "Find flights from JFK to CDG next month" - Specific route with date range ✓ "Fly from Los Angeles to Tokyo for a week in December" - Uses departure_date_ranges + stay_days ✓ "Paris to Barcelona for a weekend in April" - Specific route ✓ "Cheapest flight from ORD to LHR under $600" - Specific route with budget ✓ "Direct flight in business class from New York to London" - Specific route with preferences Flow: flight_calendar → flight_search (price-check) → trip → checkout The result carries search_criteria (what was actually searched) and next (the call to make next — on an empty answer up to three exact-date flight_search searches, on a non-empty one price_check for the best candidates). Follow next before telling the user there are no flights or quoting a cached price. Every price this tool returns is one adult's cached fare (search_criteria.priced_for: "one_adult"); the tool takes no party. To price the whole party, price-check the offer using flight_search with price_check and the travellers' counts in traveler_counts. Cost: 1 credit per call.

flight_exchange_commit

ChatGPT
Carries out the exchange the customer agreed to after seeing the price from flight_exchange_price: its ticket values, change fee, amount due or refunded and the new fare's terms. Call it only after that explicit approval. Guest verification: booking_ref is the Jinko JNK-XXXXXX reference (not an airline record locator or a hotel confirmation number) and last_name is the primary traveler's last name; no login is required. Call get_booking first, choose the items[] entry to service and pass its booked_item_id. Use the same booking_ref + booked_item_id + last_name on every step; booked_item_id stays the same after an exchange. Pass the same offer_id and the quote_id flight_exchange_price returned. An exchange can involve an additional payment or a pending refund. After a price refused because the exchange can't be refunded automatically there is no quote_id: committing the same offer_id without one sends the request to Jinko support. The airline reference in the answer is the locator usable on the airline's site or at check-in. If absent from this result, get_booking or the original confirmation can supply it. Next step: call flight_exchange_status to follow the exchange. Cost: 1 credit per call.

flight_exchange_price

ChatGPT
Prices one exchange offer exactly: the current ticket, the new ticket's value, the airline change fee if any, and the amount due or refunded, with the new fare's terms. It changes nothing. Guest verification: booking_ref is the Jinko JNK-XXXXXX reference (not an airline record locator or a hotel confirmation number) and last_name is the primary traveler's last name; no login is required. Call get_booking first, choose the items[] entry to service and pass its booked_item_id. Use the same booking_ref + booked_item_id + last_name on every step; booked_item_id stays the same after an exchange. Pass the offer_id from flight_exchange_shop. The answer carries the quote_id flight_exchange_commit needs. A price refused because the exchange can't be refunded automatically returns no quote_id: if the traveler confirms, call flight_exchange_commit with the same offer_id and no quote_id, which sends the request to Jinko support. When the answer carries customer_amounts (the price prints "Current ticket" and "New ticket"), its figures obey: current ticket + amount due − amount refunded = new ticket + change fee + residual value (an absent figure counts as 0). Answers without customer_amounts carry older supplier figures that need not add up; the identity does not apply to those answers. The "You pay" / "You get back" line (on older answers, "Total Due" / "Total Refund") gives the amount due or refunded directly, without reconstruction from the ticket values and change fee. A quoted refund is pending, rather than money already returned. In the structured result, customer_amounts holds the traveler's figures when present; the top-level fare_difference, original_fare, new_fare and penalty_amount keep an older supplier meaning and are not the traveler's figures. An exchange re-fares the ticket, so the traveler ends up on a different product. A price output that opens with "FARE CHANGE" states the one change the amount due or refunded does not show: a cheaper exchange can still move the traveler off a refundable fare. The "New fare" block gives the new fare's refund policy, change policy, checked baggage and carry-on. "Not stated by the fare" means the provider stated no rule: it is unknown, which is neither "non-refundable" nor "included", and the replaced ticket's rules are not carried over to it. Next step: show the customer the exact figures and the new fare's terms and obtain explicit approval, then call flight_exchange_commit. Cost: 1 credit per call.

flight_exchange_shop

ChatGPT
Lists the flights a booked flight can be exchanged for. It changes nothing. Guest verification: booking_ref is the Jinko JNK-XXXXXX reference (not an airline record locator or a hotel confirmation number) and last_name is the primary traveler's last name; no login is required. Call get_booking first, choose the items[] entry to service and pass its booked_item_id. Use the same booking_ref + booked_item_id + last_name on every step; booked_item_id stays the same after an exchange. segments_to_exchange names the legs to change, one entry per leg: the leg's first departure airport and final arrival airport as get_booking lists them, and its new departure date. Legs not listed stay exactly as booked. Each exchange offer in the answer carries an offer_id. Next step: call flight_exchange_price with the offer_id the customer chose. Cost: 1 credit per call.

flight_exchange_status

ChatGPT
Reports the progress of an exchange started with flight_exchange_commit. It changes nothing. Guest verification: booking_ref is the Jinko JNK-XXXXXX reference (not an airline record locator or a hotel confirmation number) and last_name is the primary traveler's last name; no login is required. Call get_booking first, choose the items[] entry to service and pass its booked_item_id. Use the same booking_ref + booked_item_id + last_name on every step; booked_item_id stays the same after an exchange. The same booking_ref + booked_item_id identifies the booked item after an exchange. Cost: 1 credit per call.

flight_price_advice

ChatGPT
Accepts a flight with its current price and returns price advice when data covers the request. The response contains fixed 1-day and 3-day price forecasts; the user does not select a wait duration. Each available forecast includes a forecasted price and estimated probabilities that the price will increase, decrease, or remain unchanged relative to current_price. price_assessment compares the supplied fare with typical prices for comparable flights. recommendation.summary is deterministic and can be displayed without an LLM. The current_price is the one-adult total including tax and excluding optional paid extras. Its value is an integer in minor units: {value:32000,currency:"USD",decimal_places:2} means USD 320.00. Initial numeric coverage is one-adult, nonstop, USD; one-way, and round-trip with return_date. A round trip requires return_date, on or after departure_date; omit return_date for a one-way trip. For a round trip, stops 0 means nonstop on both the outbound and the inbound flight. Other structurally valid currencies and stop counts are accepted and return an unavailable reason when statistics do not cover them. Omit stops when the actual number of connections is unknown. This tool does not search, monitor, quote, or book flights. Cost: 0 credits per call.

flight_refund_check

ChatGPT
Read-only eligibility probe for cancelling a booked flight. It quotes no customer figure and returns no quote handle, so nothing can be committed from it: use flight_refund_preview for the amount the customer gets back. Guest verification: booking_ref is the Jinko JNK-XXXXXX reference (not an airline record locator or a hotel confirmation number) and last_name is the primary traveler's last name; no login is required. Call get_booking first, choose the items[] entry to service and pass its booked_item_id. Use the same booking_ref + booked_item_id + last_name on every step; booked_item_id stays the same after an exchange. A "Reason:" line on a check or a refusal is the supplier's classification (NON_REFUNDABLE_FARE, EXCHANGED_TICKET, CARRIER_NOT_AUTHORIZED, DOCUMENT_RETIRED). These distinguish a fare nobody can refund from a ticket requiring an agent's manual refund; they are separate outcomes rather than a single "not refundable" answer. Cost: 1 credit per call.

flight_refund_commit

ChatGPT
Cancels a booked flight at the figures flight_refund_preview quoted. Call it only after the customer has seen that preview and explicitly approved it. Guest verification: booking_ref is the Jinko JNK-XXXXXX reference (not an airline record locator or a hotel confirmation number) and last_name is the primary traveler's last name; no login is required. Call get_booking first, choose the items[] entry to service and pass its booked_item_id. Use the same booking_ref + booked_item_id + last_name on every step; booked_item_id stays the same after an exchange. Pass quote_id = the "svq_…" handle from the preview and acknowledged = the preview's refund object (value, currency, decimal_places) copied field for field. It returns an "svc_…" operation handle. The platform refuses a commit without a live quote handle from a preview. When the amount has moved since the preview, it refuses the commit and returns a fresh quote with the new figure: show that figure to the customer before committing again. For a preview with support_level "MANUAL_REQUIRED", manual_ok: true hands the cancellation to Jinko support; nothing reaches the airline until a person at Jinko completes it. Such a commit includes acknowledged when the preview carries a refund. When the penalty could not be established and the preview carries no refund, manual_ok travels without acknowledged and Jinko support establishes the amount. Next step: call flight_refund_status with the operation handle. Cost: 1 credit per call.

flight_refund_preview

ChatGPT
Prices the cancellation of a booked flight. It cancels nothing and costs the customer nothing. Guest verification: booking_ref is the Jinko JNK-XXXXXX reference (not an airline record locator or a hotel confirmation number) and last_name is the primary traveler's last name; no login is required. Call get_booking first, choose the items[] entry to service and pass its booked_item_id. Use the same booking_ref + booked_item_id + last_name on every step; booked_item_id stays the same after an exchange. Returns whether this is a refund or a void, the penalty, the customer refund, the covered tickets, and an "svq_…" quote handle. A void returns the original charge except any part the preview lists as paid later; a refund returns what the customer paid less the penalty. The preview's penalty and customer refund are the figures the customer agrees to; it also lists the covered documents and marks extras excluded from its immediate refund. fee_known: false means the penalty and the amount remain to be confirmed, rather than free cancellation. support_level "MANUAL_REQUIRED" means a person at Jinko completes the cancellation. Next step: show the customer the figures and obtain explicit approval, then call flight_refund_commit with the quote handle as quote_id. Cost: 1 credit per call.

flight_refund_status

ChatGPT
Reports where a flight cancellation started with flight_refund_commit got to: the airline's answer, the state of every ticket, and the progress of the refund through its terminal state. It changes nothing. Guest verification: booking_ref is the Jinko JNK-XXXXXX reference (not an airline record locator or a hotel confirmation number) and last_name is the primary traveler's last name; no login is required. Call get_booking first, choose the items[] entry to service and pass its booked_item_id. Use the same booking_ref + booked_item_id + last_name on every step; booked_item_id stays the same after an exchange. Pass operation = the "svc_…" handle flight_refund_commit returned. Cost: 1 credit per call.

get_best_price_guarantee

ChatGPT
Best Price Guarantee: reads the result of a claim already made with claim_best_price_guarantee. It changes nothing, and it does not make a claim: to ask for a lower price, call claim_best_price_guarantee. HOW TO CALL: - id = the "ngo_…" id the claim returned (not a trip id, not an offer id). - wait_seconds (0–20, default 15) is how long the call waits for a review that is still running. WHAT THE TRAVELLER MUST SUPPLY: - The link to the page showing the lower price — the hotel's own page on that site, for the same dates and guests — and, on most accounts, an https address a screenshot of that page can be downloaded from. Both come from the traveller. Never build, guess or reuse one. Jinko opens the link and checks the hotel, the dates, the guests, the room, the board, the cancellation terms and the price. - The total they would pay at booking on that page, in the currency of the Jinko offer. Taxes payable at the property are not part of it. A price in another currency is declined, not converted. WHAT COMES BACK: - "offered": a new offer (offer.trip_item_token, starting with htl_neg_) at offer.price_money, valid until offer.expires_at. offer.saving_money is the saving against the rate's current standard price (offer.original.total_money, which can be lower than the price the earlier search showed); 0 means that current price was already at or below the reference. - "declined": no offer, with decline.reason, and decline.detail when the check of the reference page found something to explain — pass it on to the traveller. The offer from hotel_search can usually still be booked at its own price. - "pending": the review is still running; reading the reference page can take a few minutes. It is not a decline and not a failure — read the result with get_best_price_guarantee; do not tell the traveller it failed. - "expired": the offer was not used in time. BOOKING THE NEW OFFER: - Call trip with add_item.trip_item_token = offer.trip_item_token, then checkout, exactly as for any hotel offer. - A trip that holds the new offer can hold nothing else. Start a new trip (no trip_id); or, when the trip holds only the offer being replaced, send remove_item for that item and add_item for the new offer in the same trip call. Adding it beside another item is refused. - The price at checkout can move with the supplier. WORDING: this is the price Jinko can offer after its own review. Do not tell the traveller that Jinko negotiated with the hotel or contacted it. A claim can be declined, so do not promise a match before the result says "offered". Cost: 0 credits per call.

get_booking

ChatGPT
Start here before flight_refund_preview, flight_refund_check, flight_exchange_shop, hotel_cancel_preview or car_cancel_preview. Retrieve a booking by its Jinko booking_ref, then choose the intended items[] entry. Each item has a stable booked_item_id (an "itm_…" string), itinerary summary, confirmation, supplier_references (the confirmations the airline, hotel or rental company issued), status and servicing eligibility: can_refund, can_void, can_exchange, refund_support_level, exchange_support_level and refusal reasons/reason_code. Use booking_ref + that booked_item_id on every subsequent servicing action. Keep the same booked_item_id after an exchange. booking_state summarizes the booking's progress after payment. A flight item's confirmation.details.flight.fare_conditions lists the fare's refund and change rules (rule_type, penalty, penalty_fee). They are an estimate kept with the booking, not a live check: whether a refund or change can be done is confirmed by the refund or exchange tools when it starts. check_eligibility (default true) checks each item's servicing eligibility live with the airline or supplier, which can add several seconds for a flight. Send check_eligibility: false to list or display a booking; leave it out or send true when the user wants to cancel or change, before the refund or exchange tools. last_name is required for guests and optional for an authenticated owner. No login is needed for guest lookup. USAGE: get_booking({ booking_ref: "JNK-A7B3X9", last_name: "Doe" }) booking_ref is the JNK-XXXXXX reference from the confirmation email, get_trip or submit_agent_payment. The airline record locator (pnr) in a flight item's supplier_references is for airline check-in; servicing uses booking_ref + booked_item_id. Cost: 1 credit per call.

get_trip

ChatGPT
Fetch the full lifecycle state of a trip — its items, travelers, quote status, fulfillment status, and any booking references already produced. USAGE: get_trip({ trip_id }) This is a read-only call. Use it to inspect a trip after trip(add_item), trip(upsert_travelers), or checkout — for example to confirm what items / travelers are currently on the trip, or to check whether a booking has been finalized post-checkout. In structuredContent, lifecycle data is nested under trip. Until payment starts, trip.status reads "draft": checkout prices the trip but books and charges nothing, so a priced, unpaid trip reads trip.status "draft" with trip.quote.status "completed". Read progress from three fields: trip.quote.status (pricing), trip.payment_attempt.payment_status (payment), trip.booking_state (booking: awaiting_payment, processing, completed, partial, failed, cancelled or needs_review; trip.fulfillment.status has the detailed status). trip.fulfillment, trip.booking_ref and trip.booking_state appear when payment starts (the traveller presses Pay at checkout_url, or submit_agent_payment is called); trip.booking_state "awaiting_payment" means the payment is not authorized yet. The text answer's Stage line says where an unpaid trip stands. Read trip.booking_ref (JNK-XXXXXX) for the Jinko booking reference, which get_booking, the flight_refund_, flight_exchange_ and hotel_cancel_ tools and the car_cancel_ tools take as booking_ref. It can be present before payment completes — do not treat its presence as paid; check trip.booking_state. The value never changes afterwards. trip.bookings[].supplier_references are the SUPPLIER confirmations (airline record locator / hotel confirmation number, each with its issuer) and are NOT accepted by get_booking. trip.quote.invalid_reason = "travelers_changed" means the travelers were edited after the quote was priced: the quote can no longer be paid, so call checkout again before paying. trip_link (on a trip holding a flight) is the Jinko page where every traveler is entered and the trip is paid. travelers_on_jinko_page true means the flight is priced for a child or an infant: give the user trip_link and collect no traveler details for that trip. A traveler under 13 is shown with traveler_ref and traveler_type only. Cost: 1 credit per call.

hotel_cancel_commit

ChatGPT
Cancels a booked hotel stay at the figures hotel_cancel_preview quoted. Call it only after the customer has seen the penalty and the refund from that preview and explicitly approved them. Guest verification: booking_ref is the Jinko JNK-XXXXXX reference (not an airline record locator or a hotel confirmation number) and last_name is the primary traveler's last name; no login is required. Call get_booking first, choose the items[] entry to service and pass its booked_item_id. Use the same booking_ref + booked_item_id + last_name on every step; booked_item_id stays the same after an exchange. Pass quote_id = the "svq_…" handle from the preview and acknowledged = the preview's refund object (value, currency, decimal_places) copied verbatim, never recomputed. It returns an "svc_…" operation handle. If the refund has moved since the preview, the commit is refused and a fresh quote is returned: show that one to the customer rather than committing the old figure. Next step: call hotel_cancel_status with the operation handle. Cost: 1 credit per call.

hotel_cancel_preview

ChatGPT
Prices the cancellation of a booked hotel stay. It cancels nothing. Guest verification: booking_ref is the Jinko JNK-XXXXXX reference (not an airline record locator or a hotel confirmation number) and last_name is the primary traveler's last name; no login is required. Call get_booking first, choose the items[] entry to service and pass its booked_item_id. Use the same booking_ref + booked_item_id + last_name on every step; booked_item_id stays the same after an exchange. Returns the penalty, the customer refund, the cancellation policy, and an "svq_…" quote handle. When the preview reports the penalty is not known (fee_known: false), tell the customer we will confirm the amount: that is not free cancellation. Next step: show the customer the penalty and the refund and obtain explicit approval, then call hotel_cancel_commit with the quote handle as quote_id. Cost: 1 credit per call.

hotel_cancel_status

ChatGPT
Reports where a hotel cancellation started with hotel_cancel_commit got to: the supplier's answer and the progress of the refund. It changes nothing. Poll it until the cancellation and the refund reach a terminal state. Guest verification: booking_ref is the Jinko JNK-XXXXXX reference (not an airline record locator or a hotel confirmation number) and last_name is the primary traveler's last name; no login is required. Call get_booking first, choose the items[] entry to service and pass its booked_item_id. Use the same booking_ref + booked_item_id + last_name on every step; booked_item_id stays the same after an exchange. Pass operation = the "svc_…" handle hotel_cancel_commit returned. Cost: 1 credit per call.

hotel_details

ChatGPT
Fetch rich metadata (gallery, facilities, policies, per-room metadata) for a single hotel. Use this AFTER hotel_search returns a list of hotels and the user wants a closer look at one. REQUIRED: - hotel_id: the hotel_id of a hotel from hotel_search, passed exactly as returned OPTIONAL: - checkin, checkout (YYYY-MM-DD): accepted; the platform does not use them yet. RESPONSE: - hotel: { id, name, star_rating, address, images[], facilities[], policies[] } - rooms[]: per-room metadata, each with name + description, and max_adults / max_children (when available; max_children 0 means the room takes no children) Notes: - Read-only. Does not affect the trip or quote. - Property policies and room metadata are not rate cancellation terms. Read refundability and the cancellation schedule from the selected hotel_search rate. - Independent of pricing — call hotel_search first to get live rates with offer tokens. Cost: 1 credit per call.

price_monitoring

ChatGPT
Cache-only flight price snapshot for a fixed origin/destination + dates pair. Returns one snapshot per call. The calling agent or application must schedule repeated calls and send notifications; this tool creates no persistent monitor or alert subscription. Intended for scheduled polling — NOT for one-shot shopping or booking. WHEN TO USE: - Track the price of a specific route + dates over time and react when it drops (cron / scheduled job). - Recommended cadence: poll no faster than the cache refresh interval. Polling faster is wasted work. WHAT IT RETURNS: - The single cheapest cached itinerary matching the request filters (lowest WITHIN the filter constraints — filters constrain the candidate set, not the win condition). - Its price is a ONE-ADULT fare. traveler_counts has no effect on the fare, max_total cap, or returned flight (the price and cap are for one adult; the counts are only echoed back in request_echo); the response contains no booking link. Use flight_search with max_total for the exact whole-trip budget check for all travelers. - An offer_token that remains re-shoppable via flight_search (price_check mode) when the user decides to book. - status: "ok" (offer present) or "stale" (cache had no matching itinerary — retry later, or call flight_search for a live quote). WHEN TO USE flight_search INSTEAD: - The user is ready to book NOW and needs a live quote. - One-shot exploration where you don't intend to come back. WHEN TO USE flight_calendar / find_destination INSTEAD: - Flexible dates or ranges — price_monitoring requires exact departure_date (+ optional exact return_date). - The user is still exploring and hasn't committed to a specific route to monitor. CACHE-ONLY CONTRACT: - This tool NEVER triggers a live connector call. status:"stale" is the correct response when the cache is cold — do not retry in a tight loop and do not silently fall back to flight_search. INPUT NOTES: - Origin / destination: the city code when the user named a city (PAR, NYC, LON); origin_type / destination_type "airport" when they named an airport. Untyped, a code that names a multi-airport city is the city and any other code is the airport — LAX, MIA and LAS are read as the city. - Use the SAME filter set the user has been monitoring against between polls. The result carries search_criteria (what was actually monitored). On status: "stale" it also carries next: the live flight_search for the same dates, to offer when the user needs a price now — monitoring itself stays cache-only. Cost: 1 credit per call.

submit_agent_payment

ChatGPT
Submit a Shared Payment Token (SPT) to schedule the agent fulfillment and authorize the booking server-side in one call. This is the agent-pay alternative to sending the user to a checkout page. PREREQUISITE: call checkout first (and again if travelers were edited since, or the quote expired and checkout did not say payment stays open — such a quote is refused with 410 QUOTE_EXPIRED and nothing is charged). Use this tool only when checkout returned agent_spt_params, then mint an SPT scoped to those params; otherwise use checkout_url. USAGE: submit_agent_payment({ trip_id, shared_payment_token }) submit_agent_payment({ trip_id, shared_payment_token, quote_id }) To bind the payment to the quote the customer approved, pass the quote_id checkout returned. A quote_id that is not one of the trip's quotes is refused and nothing is charged; calling again with the same trip_id, quote_id and token returns the first attempt's outcome instead of charging again. Without quote_id, the trip's latest valid quote is charged and the answer names it as quote_id: if checkout ran again after the customer saw the price, that is the newer quote. On success the booking proceeds to fulfillment ({ status, payment_verified: true, booking_ref }). Keep booking_ref (JNK-XXXXXX) as the Jinko booking reference; get_trip(trip_id) returns the same booking_ref if this reply is lost. Its presence does not prove payment — read payment_verified and fulfillment status. For guest servicing, ask for the traveler's last_name and pass booking_ref + last_name to get_booking, the flight_refund_, flight_exchange_ and hotel_cancel_ tools or the car_cancel_ tools; booking_ref is not a supplier PNR or confirmation number. If the issuer requires customer authentication (3DS), the response carries payment_status: requires_authentication, an authentication object ({ url, expires_at }) and no checkout_url — read your issued token's next_action.use_stripe_sdk.value with your Stripe secret key and give the traveler authentication.url + "#hv=" + that value (unchanged) to complete the bank challenge before quote_expires_at, or run stripe.handleNextAction({ hashedValue }) on your own page, then poll get_trip until payment_attempt.payment_status is none of requires_authentication, cancelling or pending (cancelling is a transient cleanup state and pending is still in progress — keep polling): authorized or captured means the payment went through and fulfillment proceeds (watch booking_state); failed, declined or cancelled means it did not — submit again with a new token before the deadline; expired means the quote expired with it — call checkout again to price a new quote, then mint and submit a new token. If the card is declined, the response carries a checkout_url instead of booking_ref — open it in a browser to pay with another method. Cost: 1 credit per call.

trip

ChatGPT
Create and manage a trip: add flight, hotel, or car items, remove items, set traveler details, and select ancillaries. Operations: - add_item: Add a flight, hotel, or car item to the trip. Pass trip_item_token from flight_search (one per fare), hotel_search (htl_ tokens) or car_search. Omit trip_id to create a new trip. MULTI-ROOM HOTEL: to book several rooms of the SAME hotel and stay as ONE booking with ONE reference, make ONE add_item call with rooms: [{trip_item_token: "htl_..."}, ...] (2–8 entries, one htl_ rate token per room, mutually exclusive with trip_item_token; not every hotel rate can be booked this way) — do NOT add the same hotel twice as separate items. - remove_item: Remove an existing item from the trip by item_id (from trip response items[].item_id). Requires an existing trip_id. Can be combined with add_item in a single call to swap an item — remove runs first, then add. - upsert_travelers: Set traveler details and contact info on an existing trip_id. - select_ancillaries: Select bags, meals for a quoted trip item. Requires trip_id and item_id. Returns trip_id for use with the checkout tool. Flights, hotels, and cars can coexist in the same trip (single checkout). IMPORTANT — TRAVELER DATA: NEVER invent or fabricate traveler data, and NEVER ask for the name, date of birth or any other detail of a traveler under 13. - A flight searched with children or infants: do not call upsert_travelers. trip and get_trip return trip_link — give it to the user. On that Jinko page they enter every traveler and pay. - Anyone else under 13 in the party (for example a 12-year-old counted as an adult in the flight search): give the user trip_link the same way. upsert_travelers refuses a traveler under 13, saves nothing, and answers with trip_link. - Otherwise ask the user, for each traveler (all aged 13 or older): first_name, last_name, date_of_birth (YYYY-MM-DD), and gender (MALE/FEMALE) when the trip contains a flight. traveler_type is adult. - Hotels (a trip with no flight): enter only the guests aged 18 or older — a hotel counts anyone younger as a child. Guests under 18 are given only as children_ages in hotel_search, never as travelers. - Contact: email, phone in international format with a leading + and the country code (e.g. +12025550147); a number without the + is rejected with INVALID_PHONE_NUMBER - After adding a flight, send upsert_travelers in a call of its own: whether the flight carries a child decides whether travelers are entered here. If a required field is missing, ASK the user. Do not use placeholder or default values. ANCILLARY FLOW: Ancillaries (bags) appear as available_ancillaries on trip items once checkout has quoted the trip; offer them to the user before they pay. Use select_ancillaries with offer_ids from available_ancillaries. Full replacement — send every desired selection (bags) in each call. Flows: - Flights: flight_calendar → flight_search → trip(add_item) → trip(upsert_travelers) → [checkout → trip(select_ancillaries)] → checkout - Flights with a child, an infant or anyone under 13: flight_search → trip(add_item) → give the user trip_link (travelers and payment on that Jinko page) - Hotels: hotel_search → trip(add_item) → trip(upsert_travelers) → checkout - Cars: car_search → trip(add_item) → trip(upsert_travelers) → checkout - Mixed: supported item kinds in the same trip, one checkout Cost: 1 credit per call.

Capabilities

Writes

Example Prompts

Click any prompt to copy it.

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/jinko"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/jinko)