Slipstream: short term rental analytics

rentals.slipstream由 slipstream.rentals 發布官方發布者v1.0.1更新於 Nov 6, 2026

Read-only analytics for short-term rentals: bookings, occupancy, forecasts, reviews, owner payouts.

已驗證Streamable HTTP可網頁執行資料與分析商業與電商

服務詳細資訊

識別碼: rentals.slipstream/slipstream
傳輸: Streamable HTTP
執行環境: Web executable
信任: Verified
發布於: Oct 30, 2026

安全提示

MCP 服務會接收對話中的工具呼叫請求，並可能執行外部操作。啟用前請先了解該服務能做什麼，並只授予它所需的權限。

註冊中心已驗證此服務的命名空間屬於 slipstream.rentals，且該網域同時執行其端點。這可以確認發布者，但並不代表該服務的工具是安全的。

SourceWeft 的檢查只涵蓋條目是否完整、端點是否回應；不審查伺服器的程式碼，也不審查其工具的行為。

在 README 中新增徽章

向讀者展示該 MCP 服務已收錄於 SourceWeft，徽章連結到本頁。

[![Listed on SourceWeft](https://sourceweft.com/badge/mcp/rentals.slipstream%2Fslipstream.svg)](https://sourceweft.com/mcp/rentals.slipstream%2Fslipstream?utm_source=readme&utm_medium=badge)

概覽

AI 產生的概覽

為短租房東提供的唯讀分析：訂房、住房率、預測、評價與屋主結算，並附回原始頁面的連結。

可以做什麼: 查看資料新鮮度、列出房源，讀取房東目標與依優先順序排列的下一步動作
彙總訂房、住房率、空置夜、房源設定與未來營收預測
依類別讀取評價、房源亮點與各房源評價概況
檢查金流問題、屋主撥款、每次入住成本，並將屋主對帳單與訂房核對
解釋與說明已啟用配方的工作方式、儲存輸出與歷史命中紀錄
為某房源產生回訪旅客優惠草稿
適用情境: 適合希望讓助理直接回答自家房源組合問題的短租房東、共同房東與管理者，例如住房率、定價缺口、評價趨勢、撥款與對帳單差異，不必匯出試算表。
執行需求: 遠端 streamable HTTP 伺服器，使用公開的 Slipstream 網址，無需本機安裝。需以 Slipstream 帳號（試用或付費）透過 OAuth 登入；共同房東與管理者帳號可連線，屋主帳號目前不支援。未登入可查看工具清單，但每次呼叫工具都需已登入帳號。

安裝前請注意

登入後助理可讀取房東經 Hospitable、OwnerRez 或 Hospitable Connect 接上的 Airbnb 等短租資料。工具皆為唯讀，不能變更房源、價格或設定，也不能發送訊息，且不會回傳旅客姓名、電子郵件與電話。呼叫紀錄保留 90 天，可在 Slipstream 設定中檢視或中斷連線。

安裝

在 SourceWeft 中

開啟儀表板中的 Slipstream: short term rental analytics，將其新增到工作區。
連接所需的憑證。它們會被私密儲存，絕不會在公開頁面上展示。
為需要使用其工具的對話啟用該服務。

Web executable，透過 Streamable HTTP。遠端服務在工作區中設定後即可從網頁執行環境執行。

交給代理程式安裝

把這段提示詞貼上到 Claude Code、Codex、Cursor 等代理程式，或 SourceWeft 對話中。代理程式會讀取該伺服器的安裝指南，先向你展示發布者、端點和登入方式，經你同意後再把它新增到你的客戶端。金鑰不會寫進任何命令。

閱讀 https://sourceweft.com/mcp/rentals.slipstream%2Fslipstream/install.md，並按其中說明把這個 MCP 伺服器接入我的客戶端。修改任何設定之前，先告訴我它的發布者、端點和登入方式，等我確認。不要把 API 金鑰寫進命令或我會提交的檔案；問我金鑰應從哪裡取得。

檢視代理程式所依據的安裝指南

其他 MCP 客戶端

把它新增到你客戶端的 mcpServers 設定中。客戶端會在首次連接時完成 OAuth 登入。

{
  "mcpServers": {
    "slipstream-short-term-rental-analytics": {
      "type": "http",
      "url": "https://slipstream.rentals/api/ask/mcp"
    }
  }
}

README

Slipstream MCP server

Slipstream connects AI apps to a short-term rental host's own data: bookings, occupancy, forecasts, reviews and owner payouts from Hospitable, OwnerRez or an Airbnb account connected through Hospitable Connect. Every tool is read-only, and every figure comes with a link to the Slipstream page it came from.

This repository holds the server's public metadata and test prompts. The server itself is hosted; there is nothing to install or run.

Server URL: https://slipstream.rentals/api/ask/mcp (streamable HTTP)
Sign-in: OAuth with a Slipstream account (trial or paid). Co-host and manager accounts can connect; owner accounts can't yet.
Registry name: rentals.slipstream/slipstream (server.json)
Product page: https://slipstream.rentals/integrations/claude?ref=github

Add it

Claude: open Slipstream in Claude's connector directory and choose Connect. Or, in Claude, open Customize → Connectors → Add custom connector, and paste the server URL.

Claude Code:

Any MCP client that supports remote servers with OAuth sign-in can connect the same way. initialize and tools/list answer without a login, so you can see the tools first; every tools/call needs a signed-in account.

Tools

Tool	What it reads
`get_data_status`	How fresh the data is: when the last full sync with the PMS completed, and whether its bookings and reviews fetched cleanly.
`list_properties`	The properties this viewer can see, with their ids.
`get_goals`	The host's goals: each goal's sentence, current reading, target and gap.
`get_next_moves`	The ranked list of moves that matter most now: money errors first, then open nights, then Top Priorities, then goals.
`get_reservations_summary`	Booked stays grouped by check-in month, by channel, or by property: count, nights, revenue and average lead time (days from booking to check-in), then cancelled bookings and their revenue over the same span and groups.
`get_occupancy`	Occupancy for one or more calendar months, per property and for the portfolio: nights booked within those months (stays crossing a month's edge counted only for their nights inside it), nights in the period, and the open nights left.
`get_open_nights`	Unbooked nights right after a checkout in the next 14 days, per property, ranked by what they could earn (nights × recent nightly rate), with the minimum stay that limits selling them.
`get_listing_settings`	The pricing and availability settings last read from the PMS for each property: base and weekend nightly price, minimum stay, preparation time, cleaning fee per channel, and settings changes in the last 90 days.
`get_forecast`	Expected revenue for months from now on, with what is booked, what usually still comes in, the same month last year, the range, and how often that range held in past forecasts.
`get_metric`	One named metric for each property (or one property).
`get_recipe_output`	The latest stored outputs of a docked recipe (Top Priorities, Booking Pace Sentinel, The Pricing Oracle, The Maintenance Prophet, Reviews Owed, Fee Gaps and the rest) for each property, with when each was computed and what data the recipe reads.
`get_review_summary`	Reviews per property: share of eligible stays reviewed (with its 95% range and how it compares with the account), stays still owed a review, the rating trend, and the complaint group costing the most stars.
`get_review_categories`	Reviews by category per property (cleanliness, broken items, noise, check-in, Wi-Fi, value and the rest): the stars per rated review that guests' complaints about each category cost, how many reviews complain about it and how many praise it, plus the categories guests praise most.
`get_listing_highlights`	What one property offers that a listing title or description could lead with: bedrooms, guests it sleeps, property type and town; its most valuable amenities; amenities it has that the listing text never mentions or mentions only in passing; and what guests praise most in reviews.
`get_money_checks`	Money correctness per property: bookings missing a fee the property normally charges, double bookings and duplicate records, open owner incidents (refunds and over-limit charges), whether an owner split is set, and whether statements are on file for reconciliation.
`get_owner_payouts`	What owners earn.
`get_cost_per_stay`	Consumables (supplies) and housekeeping cost per stay per property, from the owner statements on file: the latest month's figure over a trailing three months, the usual figure lately, whether the latest month is a spike against the property's own history, and the published range for comparison.
`check_statement_month`	One month's owner statement against the bookings, per property: how many bookings the statement lists, statement lines for bookings the PMS doesn't have, that month's bookings missing from the statement, bookings whose rent differs between the two, and open guest refunds from that month and whether the statement carries them.
`explain_value`	How a docked recipe computed one output for one property: each step in order, with its value and what it reads.
`track_record`	What has been checked against outcomes: each recipe's matured predictions and how many held, the changes this host logged with how each was scored, and the recipe forecasts being tracked against what happens.
`open_draft`	Open the repeat-guest offer for a property: past guests whose reviews said they'd return, matched to an email, with a draft offer the host edits and sends from the property page.
`describe_recipe`	How a docked recipe works, without running it: each step in order, what it does, the steps it reads from, and the stored fields and data sources it reads.
`check_statement_against_bookings`	Check an owner statement the user pasted or attached (from a property manager or co-host) against this account's own bookings for that property.

What it never does

It can't change a listing, a price or a setting, and it can't send messages.
Guest names, emails and phone numbers are never returned.
Reads cost nothing in Slipstream credits.
Settings → Connected AI apps in Slipstream lists each connection, when it was last used and what it read, with a Disconnect button. Calls are logged for 90 days.

Test prompts

prompts.md is the set used to check that a model picks the right tool from these descriptions, including requests it should decline.

ChatGPT plugin package

openai-plugin/ is the package for OpenAI's Plugin directory: plugin.json (listing, review test cases), mcp.json (the one server) and the icons. To build the ZIP:

cd openai-plugin && zip -X -r ../dist/slipstream-openai-plugin-1.0.0.zip plugin.json mcp.json assets

cd openai-plugin && zip -X -r ../dist/slipstream-openai-plugin-1.0.0.zip plugin.json mcp.json assets

mcp.json holds only the server URL: the Agent Plugins schema rejects anything else on an HTTP server, so OAuth is chosen in the portal. No credentials go in the package; reviewer sign-in is entered in the dashboard.