Some phrasings get better answers than others, though, and the difference is not always obvious. These are prompts that work, grouped by what you are trying to get done.
Why does every prompt here say ezeep?
Naming ezeep tells the assistant which tools to reach for. AI clients often carry several connectors at once, and an unqualified request to print gives the assistant nothing to choose on, so it may pick another tool or ask you which one you meant. Put ezeep in the sentence and the ambiguity disappears.
Starting from a goal rather than a printer
The ezeep documentation tools work before you sign in, so an assistant can read how ezeep printing works and plan the integration while you are still deciding.
"I want to add printing to my CRM with ezeep so users can print a customer record. How would I do that?"
"Our warehouse app generates shipping labels as PDFs. What would it take to send them to a Zebra printer using ezeep?"
"Read the ezeep integration guide and tell me what a print button in my app actually requires."
"I'm building an invoicing tool. Add an ezeep print action that sends the invoice to a printer the user picks."
"Compare using ezeep through MCP at runtime against calling the ezeep REST API from my backend. Which fits a multi-tenant SaaS product?"
Expect a question before you get code. Any prompt that starts an ezeep integration triggers a choice between two authentication models, and the assistant has to put that choice to you rather than make it. Shared account means one ezeep identity prints for the whole application, which suits internal tools, kiosks and automation. Per-user means each of your users signs in with their own ezeep account, which is what a multi-user product needs. Answer, and the assistant continues with the right path. Build an app that prints with the ezeep API covers what follows.
"What data does my app send to ezeep when it prints, and where is it processed?"
"What does ezeep need from me before I can run a test print from my app?"
The second one is worth asking early. It saves the loop where the code gets written and then the account, printer and connector turn out not to be ready.
Getting oriented
"Which ezeep account and organization am I connected as, and what roles does it have?"
Worth running first. The answer tells you which of the prompts below will work for you.
"List all the ezeep printers I have access to."
"How many groups does my ezeep organization have in total, and list them all. Fetch every page, not just the first."
Checking printer capabilities
"Can the ezeep printer X print double-sided, and what paper sizes does it support?"
"What is the default paper size and tray on ezeep printer X?"
"Which of my ezeep printers support color?"
The last one runs several calls behind the scenes, since the printer list carries no capabilities and each printer has to be checked in turn. Fine for a handful of printers, slow across a large fleet.
Connector health
"Which of my ezeep connectors are offline right now, and when was each one last seen?"
"Give me a summary of my ezeep connectors: name, type, status and software version."
"Are any ezeep printers currently reporting issues?"
"Which ezeep connectors are running an older software version than the rest?"
Printing
"Print the PDF at [URL] to ezeep printer X, two copies, double-sided."
"Print [URL] through ezeep to printer X, in color on A4."
"I submitted an ezeep print job a minute ago. What is its status?"
Name the printer outright. A loose description leaves the assistant guessing which one you meant.
User and access management
"Create an ezeep group called X, then add printer Y to it."
"Invite [email address] to my ezeep organization."
"Which of my ezeep groups have no printers assigned?"
"Show me the ezeep users who are not in any group."
Reading the ezeep documentation
"Show me the ezeep API reference for the printing endpoints."
"Give me an ezeep code example for uploading and printing a file in Python."
"What are the ezeep API hosts and which operations live on each one?"
These work before you have signed in, because the ezeep documentation tools need no authorization.
Prompts that do not work well
"Which ezeep printers are offline?"
Printer records carry no status. Ask about ezeep connectors instead, or ask which printers are reporting issues.
"Delete the ezeep group X."
No delete tool exists. Use the ezeep admin portal.
"Cancel that ezeep print job."
Also unavailable through the MCP server.
"List all my ezeep users."
This one works, and that is the problem. It returns the whole directory in a single response, contact details included, which is more than most questions need. Ask for the specific thing you want instead.
"Print this."
Nothing identifies the file, the printer or the service. Even where ezeep is the only connector attached, the assistant has to come back and ask.
How do you get better answers?
Say ezeep, then name the printer, group or person outright rather than describing them. Ask for all pages when you want a complete list, since a partial answer looks identical to a full one. Keep compound requests for things that genuinely belong together, because asking for four unrelated things at once is where an assistant is most likely to answer two of them well and hurry the rest.
FAQs
Do I have to say ezeep in every message? Only where it removes ambiguity, which in practice is most of the time. Clients with several connectors attached need it to route the request. Once a conversation is clearly about ezeep, later messages in that same thread usually carry the context on their own.
Do I need to learn the tool names? No. Describe the outcome you want and the assistant chooses. Tool names help only when you want to be exact about which data you are asking for.
Why does the assistant sometimes give me a partial list? Groups and connectors paginate. Ask for every page, or ask for the total count so you can see whether the answer is complete.
Can I ask for several things in one message? Yes, and it works well for related steps like creating an ezeep group and assigning a printer to it. Split unrelated requests, since accuracy drops when one message asks for several unconnected things.
Why does asking about color support take so long? The ezeep printer list carries no capabilities, so the assistant reads each printer's properties in turn. Naming the printers you care about keeps it quick.
Do these prompts work the same in every AI client? The ezeep tools behave identically. Phrasing and how much the assistant does without checking back varies by client and model.