Back to Articles
Frans Veigar Garðarsson, CPO
Published
Updated
The Sweeply MCP server lets AI assistants such as Claude, Cursor and n8n read a hotel’s live housekeeping, maintenance and staffing data in Sweeply. This guide explains what you can ask, how to connect and how the beta keeps data safe.

In short
• The Sweeply MCP server connects AI assistants such as Claude, Cursor and n8n to a hotel’s live operations data in Sweeply. • During the beta it is read only, so the assistant can look at rooms, tasks, issues, staff and reports but cannot change anything. • Each API key gives access to one property, and reservations never include guest names, email addresses or phone numbers. • The server address is https://mcp.getsweeply.com, and Sweeply turns MCP on per organisation during the beta. • New tools are added from customer feedback and usually ship within days, with no reinstall needed.
The Sweeply MCP server connects AI assistants such as Claude, Cursor and n8n to your hotel’s live operations data in Sweeply. Once it is switched on, you can ask in plain language how many rooms are still dirty, which departures have no housekeeper assigned or which maintenance issues have been open for more than a week, and get the same numbers your team sees in the web app. The server is in beta and read only: the assistant can look at your property but cannot change anything yet.
Sweeply sells hotel operations software, so this article describes our own product. Every detail below comes from the Sweeply MCP documentation.
What is an MCP server, and why does a hotel need one?
MCP stands for Model Context Protocol, an open standard that lets an AI assistant call tools in other software. Instead of copying numbers out of reports and pasting them into a chat, the assistant asks the system directly and answers with live data. For a hotel, that means a general manager, rooms division manager or head of housekeeping can ask questions the way they would ask a colleague, and the assistant fetches the answer from Sweeply.
Sweeply already runs housekeeping, maintenance and guest requests on top of your PMS, with room status synced from the PMS through one of 20+ integrations. The MCP server opens that same picture to the AI tools your managers already use, without another dashboard to learn.
What can you ask the assistant?
The assistant can read your housekeeping overview, rooms, tasks, maintenance issues, staff schedule and reports. Typical questions from the documentation:
“How many rooms are still dirty right now?”
“Which departures tomorrow have no housekeeper assigned?”
“Show open maintenance issues older than a week.”
“What does next week’s workload look like?”
Because rooms and staff can be named in plain text, you can also ask about a single room or person. If a name is ambiguous, the server returns the candidates instead of guessing, so the assistant can ask you which one you meant.
What does a morning check look like?
A head of housekeeping could start the day with one short conversation. Asking for today’s overview returns arrivals, departures, stayovers and how many rooms are clean, dirty or inspected. A follow up such as “which checkouts are still unassigned?” lists those rooms by number, and “who is on shift today?” shows the team available to take them. Before the afternoon meeting, “show open maintenance issues older than a week” brings up the issues to chase, each with its room and comments.
The assistant answers each question from Sweeply’s live data, so nobody has to export a report or walk to the office to check the board. Assigning the rooms still happens in Sweeply, as the beta does not make changes.
Which tools does the Sweeply MCP server include?
The beta has twelve read tools in three groups. Assistants call get_property_context first, so they know your timezone, today’s date and your task categories before answering.
Group | Tools | What they read |
|---|---|---|
The day | get_property_context, get_day_overview, list_rooms, get_room, get_room_history, get_reservation, get_staff_today | Arrivals, departures, stayovers and cleaning status, one room’s status, stay, tasks, comments and open issues, a room’s history over up to 31 days, one stay’s dates and guest count, and who is on shift |
Tasks and issues | list_tasks, list_issues, get_issue | Tasks over up to 31 days by type, status, assignee or room, open maintenance issues by age or category, and one issue with comments and photo links |
Planning and reports | get_workload_forecast, get_housekeeping_performance | Planned arrivals, departures, stayovers and cleanings per day, and completed cleanings with average minutes by housekeeper, room or task |
Dates can be written as YYYY-MM-DD or simply as today, tomorrow or yesterday, always in the property’s timezone. Lists stop at 50 rows and say how many rows there are in total, so the assistant knows when it is looking at part of a list.
How do you connect Claude, Cursor or n8n?
Every client needs the same two things: the server address https://mcp.getsweeply.com and an Authorization header that carries your Sweeply API key. The steps:
Ask Sweeply to turn on MCP for your property. It is off by default during the beta; contact support or your account manager.
Create an API key in the Sweeply web app under Settings, Account, API Keys, and copy it straight away, as it is shown only once.
Add the server to your assistant with the address and the header Authorization: Bearer followed by your key.
In Claude on the web, desktop or mobile, the server is added under Settings, Connectors, as a custom connector with the Authorization header. Claude Code, Cursor and n8n each take the same address and header in their own settings; the Sweeply MCP documentation has the exact configuration for each. ChatGPT is not supported yet, because ChatGPT connectors sign in with OAuth and the Sweeply MCP server accepts API keys only for now.
Is the hotel’s data safe?
The design keeps the assistant inside one property and away from guest personal data:
One key, one property. The API key decides which property the assistant reads, so tools never ask for a property and cannot reach another one.
Read only. In the beta the assistant can look but cannot create, change or delete anything.
No guest personal data. Reservations carry an id, dates, room and guest count, never names, email addresses or phone numbers.
Staff text is marked as untrusted. Comments, notes and issue descriptions are flagged so the assistant treats them as information, never as instructions.
Easy to revoke. Delete a key and the assistant loses access on its next request. Give each person or tool its own key, so you can remove one without breaking the others.
One point to plan for: an API key is not a staff member, so staff roles and area groups do not narrow what it reads. A key sees the whole property. Hosting and data processing follow Sweeply’s usual setup, described on our security page.
How does it fit with Sweeply’s AI agents?
Sweeply’s AI agents work inside Sweeply: Tracey reads reservation remarks and turns them into tasks, and Roomey assigns rooms. The MCP server works the other way round. It lets the general purpose assistant your managers already use read Sweeply, so they can combine operational answers with their own email, spreadsheets or planning in one conversation.
What comes next?
The MCP server grows with customer feedback. Tell Sweeply a question the assistant could not answer, and the team replies the same day with what it plans to build; new tools usually ship within days and appear in your assistant automatically, with no reinstall and no new key. The roadmap in the documentation lists:
Making changes safely: creating and assigning tasks and reporting issues, with a personal sign in so every change is made in a staff member’s name.
Several properties at once: one connection for a hotel group.
More assistants: sign in support for ChatGPT and other OAuth only clients.
Send ideas to support@getsweeply.com or through the chat in the Sweeply app, with the question you asked and what you expected back.
Where this guide does not apply
The MCP server reads data from Sweeply, so it only helps hotels that already run their operations in Sweeply. It does not read your PMS directly beyond what Sweeply syncs, it cannot make changes during the beta, and it is not yet available in ChatGPT. If your team does not use an AI assistant at work, the reports and live overview in the Sweeply web app give the same numbers without one.
Talk to our sales team
Want to see how hotels use Sweeply, and what your managers could ask an AI assistant about your property? Book a call with our sales team.
Conclusion
The Sweeply MCP server turns your operations data into answers your managers can get from the AI assistant they already use. It is read only and in beta, connects with an API key, keeps guest personal data out, and grows with the questions hotels send in.
Frequently asked questions
What can I ask the Sweeply MCP server?
You can ask about your housekeeping overview, rooms, tasks, maintenance issues, staff on shift, workload forecasts and housekeeping performance, for example how many rooms are still dirty or which departures tomorrow have no housekeeper assigned.
Does the Sweeply MCP server work with ChatGPT?
Not yet. ChatGPT connectors sign in with OAuth, and the Sweeply MCP server accepts API keys only for now. Sign in support for ChatGPT and other OAuth only clients is on the roadmap.
Can the AI assistant change data in Sweeply?
No. During the beta the MCP server is read only. Creating and assigning tasks and reporting issues, with a personal sign in so every change is made in a staff member’s name, is planned.
How do I turn on the Sweeply MCP server?
Ask Sweeply support or your account manager to enable MCP for your property, create an API key under Settings, Account, API Keys in the web app, and add https://mcp.getsweeply.com with that key to your assistant.
Does the assistant see guest personal data?
No. Reservations carry only an id, dates, room and guest count. Guest names, email addresses and phone numbers are never returned.