WinCloud
RB Software, Inc. v0.1.0
Hotel bookings and room data
From the developer
Declared limitations
- The room features list says which features a hotel has set up, not which rooms have them. Do not present list_room_features or list_rooms output as a per-room answer; explain the limitation.
- WinCloud's tools never return payment or card details. Do not invent them or suggest another tool returns them; explain that payment details are not available through WinCloud's tools.
- WinCloud's tools can't create, change or cancel bookings. Do not claim the reservation was cancelled or call a tool as if it could; explain that the change has to be made in WinCloud itself.
Not specified in the available declarations: capabilities, pricing & account requirements.
Onboarding example
Find bookings made this month at my hotel.
Sources & dates
Developer declarations; not independently tested. Listing last checked Nov 16, 2026. Package observed Nov 15, 2026.
Description
From the marketplace listing
WinCloud brings your hotel's property management data into ChatGPT, so your front office, reservations and revenue teams can get answers in plain language instead of running reports. Ask about the bookings made, arriving, departing or cancelled over a period, look up a reservation in full by its number or find bookings by the guest's name, check the room types, rate plans and channels you sell, see nightly pricing and restrictions for the dates ahead and how each rate is derived from its master rate, list your rooms and the room features you have set up, find out how many rooms are occupied or vacant today, look up a guest's profile, contact details and stay history, and list the hotels in your WinCloud property. Booking, cancellation and no-show activity by arrival date, with lead time and average daily rate, gives you a practical view for revenue management: how bookings are coming in, how full you are, and how your pricing is set for the dates ahead. You need a WinCloud account, and what you can ask about depends on the permissions your administrator has given you. You see only the hotels your account can access.
Language: English · Automatically detected from descriptions.
Changes
No changes recorded yet. Follow this plugin to keep track of future changes.
First seen Nov 15, 2026. Last seen Nov 16, 2026.
Similar plugins
Find competitors to follow as you develop your plugin.
-
Run your hotel from ChatGPT: bookings, arrivals, availability, rates and reports in Roomnix.
Shared terms: hotel, arrival, booking, channel
-
Rio Las Vegas helps guests explore hotel details, room and suite options, images, rates, booking links, cancellation links, and service r...
Shared terms: hotel, room, booking, cancellation
-
Hotel Emperador Madrid Assistant helps guests find hotel information and complete common reservation tasks through chat. Users can ask qu...
Shared terms: booking, hotel, guest, reservation
-
Porcel hotels Assistant helps guests find hotel information and complete common reservation tasks through chat. Users can ask questions a...
Shared terms: booking, hotel, guest, reservation
-
Hotel Liabeny Assistant helps guests find hotel information and complete common reservation tasks through chat. Users can ask questions a...
Shared terms: booking, hotel, guest, reservation
Keywords
Files & skills
File archives
Skill instructions
wincloud10.7 KB
---
name: wincloud
description: Use when a user asks about their hotel's data in WinCloud, such as bookings, arrivals, departures or cancellations, one reservation in detail, booking pace and demand, room types, rooms and features, rate plans, nightly tariffs and how rates are derived, rooms occupied or vacant today, or a guest's profile and stay history. Also use for any request for stats, summaries, insights, charts, KPIs, reports, forecasts or analysis about that data. Explains which WinCloud tool answers which question, the order to call them in, and what the tools cannot do.
---
# Using WinCloud
**Every answer that summarises numbers (counts, totals, trends, comparisons, KPIs) follows `wincloud-report-format`: headline, KPI row, a real chart, then a detail table, and the user does not have to ask for the chart. When they say "all locations", use every location `list_companies` returns and show them side by side in one chart. Never answer such a question with bullets or a table alone. Lists of individual bookings or guests are the exception: a table.**
WinCloud is a hotel property management system. Its tools read one property location's (one hotel's) data at a time. They are read-only today: they never create, change or cancel anything, and they never return payment details.
## Before any call: the location
Every tool needs a `companyId`, the WinCloud code of one property location, that is one hotel (for example `WS`). WinCloud calls locations "companies", so say "hotel" or "location" to the user. The codes are internal ids: never show a bare code to the user.
1. **Resolve names first, once per conversation.** Before the first answer that mentions a hotel, call `list_companies` and keep the map of code to name (for example `WS` → Grand Hotel), with each hotel's currency and room count. Do this even when the `companyId` description already lists codes, because that list has codes only.
2. **Choosing a hotel:** if there is only one location, use it without asking. If there are several and the user didn't name one, ask with the hotel names as the options ("Grand Hotel", "Western Hotel", "Both"), not the codes, or call the tool once per location when they asked about all of them.
3. **Matching what the user says:** match a hotel the user names ("Grand", "the Western") to its code with the map. Never guess a code; the `WS` in the description is only an example.
4. If a call fails with an access error, name the hotel, say the user can't see it, and offer the ones they can, by name.
**Naming rule for every answer:** call each location by its `list_companies` name in headlines, KPI labels, chart titles, legends, axis labels, tables, attention lists and questions. The code may appear once, in brackets in "Basis and caveats" (for example "Grand Hotel (WS)"), so the manager can find it in WinCloud. If `list_companies` is unavailable or returns no name, use the code and say the name couldn't be retrieved.
## Which tool answers which question
| The user wants… | Tool |
|---|---|
| Bookings made, arriving, departing or cancelled over a period, by status, number or guest name | `search_bookings` |
| One reservation in full: rooms, times, rate plan, source, OTA reference, cancellation reason | `get_reservation` |
| Booking pace, lead times, ADR, cancellations or no-shows | `get_reservation_analytics` |
| The room types, rate types and rate plans a hotel sells, and the channels it sells through | `get_rate_plans` |
| Prices and restrictions (stop-sell, closed to arrival or departure, minimum or maximum stay) for dates | `search_room_tariffs` |
| Room types with their room counts and base prices | `list_room_types` |
| The master rates other rates are derived from | `list_master_rates` |
| How a derived rate is calculated from its master rate | `get_rate_configuration` |
| When each set of rate formulas applies (days, seasons) | `list_rate_templates` |
| The physical rooms: numbers, floors, beds | `list_rooms` |
| The room featurePreview of the first 4,000 characters. Read full instructions →
wincloud-adr-revpar3.11 KB
--- name: wincloud-adr-revpar description: Use when a user asks about average daily rate (ADR), revenue per available room (RevPAR), pricing performance, which room types or rate plans earn the most, how prices have changed over time, or how current rates compare with what guests actually paid. Works from WinCloud's reservation analytics, tariffs and room counts. --- # ADR, revenue and pricing performance Follow `wincloud-report-format`. Resolve hotel names with `list_companies` before gathering data, and label every hotel by name in the output, as `wincloud` explains. For several hotels, use one row or series per hotel name. ## ADR from bookings 1. Call `get_reservation_analytics` with `report` = `bookings`, a date range and optionally `roomTypeId`. Read all pages. 2. Each row has rooms (booked plus availed, as `wincloud` explains) and an ADR, which is the reservation's tariff as WinCloud stores it. Average ADR = Σ(ADR × rooms) ÷ Σ(rooms). Don't take a plain average of the row ADRs. Cancelled and no-show reservations are already left out. 3. Group by arrival month, room type, or meal plan for a trend or breakdown. The range filter is the **booking date**, so say that the figures cover bookings made in the range, not stays in it. For ADR of stays in a period, widen the booking range and keep only rows arriving in the period, as `wincloud` explains. 4. Rows are per reservation, not per night, so ADR here is the booking's rate, not a night-weighted figure. ## Current pricing versus bookings - `search_room_tariffs` gives the prices and restrictions set for dates. Use `get_rate_plans` first for the room type and rate type ids. Weekend tariffs are not included, so say the picture covers weekdays only. - Name the exact rate plan you compared against. For a "standard price", follow `wincloud` and never pick a plan silently. - Compare the **rate on the calendar** with **ADR on the books** for the same room type and arrival month. A large gap suggests discounted channels, packages or old rates; say it is a possible reason, not a finding. ## RevPAR, and what can't be calculated - RevPAR = room revenue ÷ available room-nights. The tools give neither room revenue nor room-nights directly, so **RevPAR cannot be exact here**. - You may give an **indicative** figure: Occupancy on the books × ADR. Label it "indicative RevPAR, based on bookings on the books, not actual revenue". Do this for future dates only, taking occupancy from `wincloud-occupancy-forecast`; past occupancy isn't available from these tools. - For actual revenue and RevPAR, tell the user to use the hotel's WinCloud reports. ## Show - **Headline:** average ADR for the period, the change against the previous period, and the top and bottom room type. - **Chart:** ADR by arrival month (line) or by room type (sorted horizontal bars). - **Table:** room type, rooms, ADR, with the calendar rate beside it when you fetched it. - **Caveats:** the hotel, the date field, pages read, the currency, and the RevPAR limit above. Analytics rows carry no currency: use the location's base currency from `list_companies`, or the currency `search_room_tariffs` returns, and say which.
wincloud-cancellations3.02 KB
--- name: wincloud-cancellations description: Use when a user asks about cancellations or no-shows, the cancellation rate, how late guests cancel, which room types or dates lose the most bookings, or how to reduce lost revenue from cancelled or unused bookings. Analyses WinCloud's cancellation, no-show and booking data. --- # Cancellations and no-shows Follow `wincloud-report-format`. Resolve hotel names with `list_companies` before gathering data, and label every hotel by name in the output, as `wincloud` explains. For several hotels, use one row or series per hotel name. ## Gather 1. **Cancellations:** `get_reservation_analytics`, `report` = `cancellations`. Its date filter is the **cancellation date**. Each row has arrival date, cancelled date, days before arrival and rooms cancelled. 2. **No-shows:** same tool, `report` = `no-shows`. Its date filter is the **arrival date**. 3. **Bookings, for a rate:** `report` = `bookings`, with its **booking date** filter. It leaves out reservations that are now Cancelled or NoShow, so it is not the total that was booked. To compute a rate: - Read the bookings and cancellations reports over a date range wide enough to cover every booking for the arrivals in question. If you can't, give counts, not a rate, and say why. - Keep only rows from all three reports that arrive in the same window. - Total = rooms on the bookings rows (booked plus availed) + rooms cancelled + no-show rooms. - Cancellation rate = rooms cancelled ÷ total; no-show rate = no-show rooms ÷ total. 4. Read every page and add the rows up. Say if you stopped early. ## Measure - **Counts:** rooms cancelled, no-show rooms, reservations affected. - **Lead time of cancellations:** bands (same day, 1–3, 4–7, 8–14, 15+ days before arrival). Late cancellations (7 days or fewer, the first three bands) are the costly ones: they leave rooms hard to resell. - **Where it concentrates:** by room type, arrival month or week. - **Reasons and sources:** `get_reservation` shows the cancellation reason and source for one booking. For a pattern, sample 5–10 recent cancellations from `search_bookings` (`cancelledFrom`/`cancelledTo`, status `Cancelled`), and say it is a sample. ## Show - **Headline:** rooms cancelled, the rate if it can be computed, and the main pattern (for example "40% were cancelled within a week of arrival"). - **Chart:** cancellations by lead-time band (bars); trend by week if the range is long. - **Table:** top room types or dates, with rooms cancelled and, where possible, rate. - **Suggestions**, as options for the manager to consider, not decisions: tighter cancellation terms for the heaviest bands, reconfirmation for late arrivals, overbooking only with the hotel's own policy. Say these depend on the hotel's contracts and policy. ## Limits - Analytics rows name no guests and no reservation numbers. - A cancellation date and an arrival date answer different questions; always say which you used. - Don't tell the user to take payment, refund or change a booking. WinCloud tools here are read-only.
wincloud-daily-briefing2.76 KB
---
name: wincloud-daily-briefing
description: Use when a user asks for today's picture at a hotel, such as a morning briefing, "how are we doing today", today's arrivals and departures, in-house guests, how full the hotel is now, or what needs attention today. Builds a one-screen summary from WinCloud's live bookings and room status.
---
# Daily briefing
Follow `wincloud-report-format`.
Resolve hotel names with `list_companies` before gathering data, and label every hotel by name in the output, as `wincloud` explains. For a briefing across hotels, repeat the steps below per hotel and show one row or series per hotel name.
## Gather (all for today's date, yyyy-MM-dd)
1. **Rooms now:** `get_inventory`: occupied and vacant by room type. Add the room types up for the total. Page through if more than one page.
2. **Arriving today:** `search_bookings` with `arrivalFrom` and `arrivalTo` = today. Read the status of each booking; count those still expected (Confirmed, Tentative, Provisional) and those already CheckedIn.
3. **Departing today:** `search_bookings` with `departureFrom` and `departureTo` = today. Count CheckedIn versus already CheckedOut.
4. **In-house:** `search_bookings` with status `CheckedIn`. Take the count from the first line of the result ("N booking(s)"); read more pages only if you need the list.
5. **Cancelled today:** `search_bookings` with `cancelledFrom` and `cancelledTo` = today.
6. **Booked today (pickup):** `search_bookings` with `startDate` and `endDate` = today.
Run independent calls together. Take counts from each result's first line ("N booking(s)") and only fetch extra pages when you list people.
## Show
- **Headline:** occupied of total rooms, arrivals and departures still to come, and the one thing that stands out.
- **KPI row:** Occupied %, Arrivals (expected / arrived), Departures (expected / left), In-house, New bookings today, Cancellations today.
- **Chart:** occupied versus vacant by room type, as horizontal bars.
- **Table:** today's arrivals, 15 rows at most: guest, room type, plan, nights, status.
- **Attention list**, only items the data supports: arrivals not yet checked in, departures not yet checked out, cancellations today, a room type with no vacancy, rooms counted as neither occupied nor vacant, and in-house bookings whose departure date has passed.
## Limits
- `get_inventory` is a snapshot of right now, so say the time you read it. It can't show tonight's or tomorrow's availability.
- Occupied % is occupied rooms divided by total rooms from `get_inventory`. If a room count comes back as "-", say it's missing; don't treat it as zero.
- Show occupied, vacant and "other or unavailable" (total minus both) so the three add up to total rooms, as `wincloud` explains.
- No payment details, and no guest contact details in a briefing.
wincloud-occupancy-forecast3.84 KB
--- name: wincloud-occupancy-forecast description: Use when a user asks how full the hotel will be, for an occupancy forecast, expected rooms or occupancy for coming days, weeks or months, which dates are weak or sold out, or how many rooms are still to sell. Estimates future occupancy from the reservations already on the books in WinCloud, optionally adding typical late bookings from the past. --- # Occupancy forecast Follow `wincloud-report-format`. This is an estimate. Say so in the headline area and again in the caveats. Resolve hotel names with `list_companies` before gathering data, and label every hotel by name in the output, as `wincloud` explains. For several hotels, use one row or series per hotel name. ## Capacity Total rooms = the sum of total rooms across room types from `get_inventory` (or `list_room_types`). `list_companies` also gives each location's room count; if the two differ, use the room-type sum and say so. Use one figure for the whole window, and say that out-of-order rooms are not known. ## Step 1: Rooms on the books, per night 1. Choose the window (default: next 30 days, at most 90). Turn it into exact dates. 2. Call `search_bookings` with `departureFrom` = window start and `arrivalTo` = window end. This returns every stay that touches the window, including ones that began earlier. Read all pages; if there are more than 15, say the forecast covers only the pages read and narrow the window. 3. Keep only Confirmed, Tentative, Provisional and CheckedIn; drop every other status (Cancelled, NoShow, TurnAway, WaitingList, CheckedOut, Unknown). 4. For each booking, count its rooms on every night from arrival up to, but not including, departure. Add them up per night. A night is occupied from the arrival date; the departure date is free. 5. On-the-books occupancy per night = rooms on the books ÷ total rooms. Mark Tentative and Provisional separately (a lighter shade): they may not convert. ## Step 2 (optional): Likely late bookings Only if the user wants a fuller forecast and you can afford the calls: 1. Do Step 1 for the same dates one year earlier (or the last comparable period), but **keep CheckedOut** as well: those nights are past, so the guests who stayed now show as CheckedOut. Count only bookings made at least as many days before each night as today is before the forecast night. That is last year's position at the same point. 2. Compare it with last year's final result for those nights (CheckedOut and CheckedIn bookings: the stays that actually happened). The gap is last year's late pickup. Bookings cancelled since then no longer count, so last year's position is slightly understated; say so. 3. Add that gap, as a share of last year's position, to this year's on-the-books figure. Cap the result at 100%. If last year is missing, too thin, or unlike this year (an event, a closure), skip Step 2 and say why. Don't invent a pickup. ## Show - **Headline:** the average occupancy for the window, the strongest and weakest dates, and the number of rooms still to sell. - **Chart:** line or bars by date: on-the-books occupancy solid; the estimate with late bookings (if you did Step 2) dashed or lighter. Mark Tentative on its own if significant. - **Table:** weeks or the 7 weakest dates with rooms on the books, rooms still to sell and occupancy. - **Caveats:** the method, the window and hotel, the pages read, and: - walk-ins and late bookings are not predicted unless Step 2 was done; - events, holidays, price changes and overbooking are not known; - the figure is an estimate to verify before it drives pricing or staffing. ## Limits - `get_inventory` is today's snapshot, so it is not a source of future availability. Future occupancy always comes from bookings. - Group blocks and allotments may not appear as bookings; the forecast can understate them. - Don't present a forecast as a promise or as the hotel's official figure.
wincloud-performance-review3.41 KB
---
name: wincloud-performance-review
description: Use when a user asks for a performance review, management or owner report, summary, dashboard, scorecard or "how did we do" for a period such as a week, month, quarter or year to date, for one hotel or all their hotels, or asks to compare hotels or compare one period with another. Builds a KPI and chart report from WinCloud bookings, cancellations, no-shows and rates, and ends with what to act on. Use it even when the user doesn't say "report" or "chart".
---
# Performance review
Follow `wincloud-report-format`. This is the report an owner or area manager reads: a few KPIs, a few charts, and what to do next.
Resolve hotel names with `list_companies` before gathering data, and label every hotel by name in the output, as `wincloud` explains. For several hotels, use one row or series per hotel name.
## Scope
- **Locations:** the hotel the user names, or every location they can see when they say "my hotels", "both" or "all". Repeat the gathering below per location.
- **Period:** exact `yyyy-MM-dd` dates; say which you used. For "versus last year" or "versus last quarter", use the same dates a year earlier or the previous period of equal length.
## Gather (per location, per period)
1. **Bookings made, by month:** `search_bookings` with `startDate` and `endDate` for each month, `pageSize` 1. Take the count from the first line ("N booking(s)"). This counts every status, by booking date.
2. **Net bookings, rate and lead time:** `get_reservation_analytics`, `report` = `bookings`. Read every page (up to 15). Each row is one reservation still on the books; rooms are booked plus availed, as `wincloud` explains. Work out rooms, ADR (weighted by rooms), lead-time bands (the bands in `wincloud-pickup-pace`) and the top room types.
3. **Cancellations:** same tool, `report` = `cancellations`. Dated by the cancellation date. The first line gives the number of reservations; for rooms, read every page and add them up.
4. **No-shows:** same tool, `report` = `no-shows`. Dated by the arrival date. Count reservations and rooms as for cancellations.
5. Say in the KPI labels whether each figure counts reservations or rooms.
6. Run independent calls together. For a rate of cancellation or no-show, use the method in `wincloud-cancellations`; without it, give counts.
## Show
- **Headline:** which hotel or period leads, by how much, and the one thing that stands out.
- **KPI row, per hotel:** bookings made, net bookings, cancellations, no-shows, ADR, average lead time.
- **Charts, one comparison each:**
- bookings made per month, hotels side by side (grouped bar);
- cancellations and no-shows per hotel, in separate charts because their date bases differ;
- top room types per hotel (horizontal bars);
- lead-time bands (bars), when you read the analytics rows.
- **Table:** one row per hotel with the KPIs, plus the comparison period when there is one.
- **What to look at:** up to 3 items the numbers support, for example a hotel whose no-shows are well above the other's, or a month that dipped.
## Limits
- Bookings made counts every status; net bookings leave out cancelled and no-show reservations. Say which one the headline uses and don't subtract one from the other.
- No revenue or payment figures come from these tools; ADR is the booking's rate, not room revenue. For revenue, point to the hotel's WinCloud reports.
- If you stopped before the last page, say the figures cover only what you read.
wincloud-pickup-pace3.02 KB
--- name: wincloud-pickup-pace description: Use when a user asks whether bookings are ahead or behind, about booking pace or pickup, how many bookings came in per day or week, lead time (how far ahead guests book), or how demand compares with the same period last year or last month. Measures booking activity from WinCloud's reservation analytics. --- # Pickup and pace Follow `wincloud-report-format`. Pace answers "are we booking faster or slower than usual?". Resolve hotel names with `list_companies` before gathering data, and label every hotel by name in the output, as `wincloud` explains. For several hotels, use one row or series per hotel name. ## Pickup: bookings made over a period 1. Call `get_reservation_analytics` with `report` = `bookings`, `startDate`/`endDate` = the booking dates the user cares about (the date filter is the booking date). Add `roomTypeId` for one room type; ids come from `get_rate_plans`. 2. Each row is one reservation still on the books (cancelled and no-show reservations are left out): booked date, arrival date, rooms (booked plus availed, as `wincloud` explains), lead time in days, ADR. Read every page and add up. Say if you stopped early. 3. Total **room-nights are not available** from these rows (one row per reservation, not per night). Report **rooms** and **reservations**, not room-nights. 4. Group by booked date for daily or weekly pickup; by arrival month for what the bookings are for; by lead-time band (booked after arrival, 0–7, 8–14, 15–30, 31–60, 61+ days) for how far ahead guests book. ## Pace: compared with another period 1. Pick two periods of equal length, the current one and the comparison (same dates last year, or the previous period). Say which. 2. Compare like with like: bookings made in each period, **for the same arrival window** (for example, rooms in the last 30 days for next January, versus the same 30 days a year earlier for the January before). Filter rows by arrival date after reading them. 3. Show the difference as a number and a percentage, and call it ahead or behind. If the base is small (under about 20 reservations), say the comparison is not reliable. ## Show - **Headline:** rooms taken in the period, the change against the comparison, and the main reason visible in the data (a lead-time shift, a room type). - **Chart:** bars for rooms per day or week, with the comparison period as a second series (green). - **Second chart or table:** lead-time bands, and top room types or plans by rooms. - **Caveats:** the hotel, both periods, the date fields used, pages read. ## Limits - The `bookings` report already leaves out reservations that are now Cancelled or NoShow, so its pickup is **net** of them. Don't subtract the `cancellations` report as well; that counts them twice. Gross pickup (including bookings later cancelled) can't be rebuilt, because cancellation rows carry no booked date. - No guest names or reservation numbers come back from analytics; use `search_bookings` for those. - Past trends don't guarantee the future. Don't call a pace figure a forecast.
wincloud-report-format7.33 KB
---
name: wincloud-report-format
description: How to present any WinCloud data so a hotel manager can read it at a glance. Use for every WinCloud answer that has more than one number or row, such as stats, statistics, KPIs, summaries, insights, analytics, reports, trends, comparisons, forecasts, "how are we doing", bookings or reservations by month, location, status or room type, even when the user never says "chart". Makes the answer a headline, KPI row, real chart and detail table, in RBS colours. The other wincloud-* skills follow it.
---
# Presenting WinCloud data
Managers skim these answers between tasks. A chart shows a pattern in a second where a table takes a minute, so draw one by default; nobody should have to ask for it. A KPI table alone is not enough: it supports a chart and never replaces it.
## Order of every answer
Every response, even a single fact, a list or "WinCloud can't do that", opens with a one-sentence summary of at most 25 words. The reader gets the answer in seconds and reads the rest only if they want the detail. For answers with numbers, that sentence is the headline below.
1. **Headline.** One or two sentences: the number, the direction and what it means ("Hotel A took 150 bookings in Q1, 100 fewer than Hotel B").
2. **KPI row.** The 3 to 6 headline figures, labelled, in one small table (labels on top, values below).
3. **Chart.** At least one real chart of the pattern, and one more per further comparison. Several locations, months or statuses means a chart that shows them side by side. Two or three charts at most; each shows one comparison, so none is crowded.
4. **Detail table.** The numbers behind the charts, 20 rows at most.
5. **Basis and caveats.** Hotel, date range, which date field was used, whether all pages were read, and anything estimated. A few lines.
6. **What to look at.** For analysis, forecast and report answers, up to 3 specific items the numbers support (a weak date, a rising no-show count). Offer them as things to consider, not decisions.
Skip the chart only for a single value or yes/no (a sentence), or a list of individual bookings, guests or rate plans (a table, newest first). If a query returns no rows, say so and name the filters you used; don't draw an empty chart.
## Pick the chart
| The data is… | Chart (`type` in `scripts/chart.py`) |
|---|---|
| Trend over dates or months | `line`; `bar` when each point is a count |
| Several locations, statuses or years compared (a vertical bar chart) | `bar`, grouped, one series per location or status |
| Categories (room types, rate plans, sources), many or long names | `hbar`, sorted high to low |
| Share of a whole, 5 parts or fewer (status mix, cancellations versus no-shows). A request for a pie chart gets this. | `donut` |
| Prices or status across dates and room types | Table with the odd cells flagged |
Figures on different date bases, such as bookings by booking date and no-shows by arrival date, go in separate charts. Never one chart that implies a single cohort.
## Drawing the chart
Work down this list and stop at the first that works. Finish the chart before writing the headline, so the words match it. Charts belong in the conversation, so show them inline where you can.
1. **An inline visualizer or chart tool, when your environment has one** (in Claude, `show_widget`, after reading its `chart` guidance with `read_me`). Pass the SVG or HTML straight to it. Don't write a file first.
2. **`scripts/chart.py`** in this skill's folder. It needs only Python. Write a small JSON spec (`type`, `title`, `subtitle`, `unit`, `categories`, `series` with `name` and `values`; `"dashed": true` on a forecast series), run `python3 <this folder>/scripts/chart.py spec.json chart.svg`, then show the picture inline (a file-sharing tool set to render in the conversation, if you have one). It writes a PNG too when it can. `references/chart-recipes.md` has the spec for each type and the next steps.
3. **Your own tool.** matplotlib, anoPreview of the first 4,000 characters. Read full instructions →
Referenced files: 2 — file references are metadata, not package contents.
Release notes
First release of WinCloud for ChatGPT. Ask about your hotel's bookings, pricing, room status and guests in plain language.
Declared in the saved package. Remote tools may change independently.
Package details
Publisher declarations from the archived package. These are separate from our research and the live service's terms.
- Package author
- RB Software, Inc.
- Keywords
- See keywords
- Declared availability
- No country restrictions declaredDeclared regions; live availability may differ.
- Commerce declaration
- Does not support commerceThis does not establish whether access is free or paid.
- Review scenarios
- 5 positive · 3 negativeScenario counts, not test results.
Package observed Nov 15, 2026.
Technical details
- First seen
- Nov 15, 2026 · 18:00 UTC
- Last seen
- Nov 16, 2026 · 18:00 UTC
- Collection status
- Collected
plugin_asdk_app_6ae8300d492c8191a072f79115188500
Download plugin data (JSON)Plugin details: freshly prepared. Prepared . Refreshed when plugin data changes or after 6 hours. Your account and Follow status are live.