This article gives examples of what support teams do with the Swifteq MCP Server, with prompts you can adapt. Change the dates, fields and wording to suit your team.
In this Article
Reporting and Analysis
Quality Assurance and Coaching
Help Center Work
Drafting Replies and Working the Queue
Curating and Applying Macros
Combining Zendesk with Other Tools
What Customers Say
Reporting and Analysis
This is the most common use of the MCP Server.
- List this week's open tickets with each customer's organization and its ARR field, highest ARR first.
- Summarize the tickets created in the last 7 days as an executive report: volume, top contact reasons, escalations and anything unusual compared with the week before.
- Build a dashboard of open tickets showing negative sentiment, overdue tickets and priority.
- What were the main reasons customers contacted us in August? Group them into no more than 10 themes with ticket counts.
- Review this month's cancellation tickets and list the customers who are at risk of leaving, with the reason.
- Find the recurring bugs reported in tickets over the last 4 weeks and rank them by number of tickets.
- What were our median first reply time and full resolution time last month, in business hours, and our one-touch rate? Compare them with the month before.
- What was our SLA achievement last week for first reply and next reply, by group? List the tickets that breached.
- Which of my open tickets have breached an SLA or are about to? Start with the most urgent.
- Show our SLA achievement for agent work time, first reply time and pausable update time for September, and the month before.
- Show a day-by-day trend of solved tickets and average resolution time for September, by group.
- How many tickets did we get from each channel in 2025?
- Count last month's tickets by Contact reason.
- What was our CSAT score last month for the Support group, and how many surveys went unanswered?
Quality Assurance and Coaching
- Review 20 tickets solved yesterday. For each, check accuracy, tone, whether one of our macros would have answered it, and whether the ticket was categorized correctly.
- Compare the CSAT ratings and resolution times of each agent this month with last month.
- Read the replies from the billing team this week and suggest three concrete ways they could communicate more clearly.
- List last week's tickets our AI agent resolved alone, and flag any where the customer didn't actually get an answer.
- How many tickets did Zendesk count as automated resolutions last week? Check which ones ended with the customer leaving without an answer.
- Read the conversation on ticket 12345. Did our AI agent answer the question, and when did a person take over?
Reviewing Zendesk AI agent conversations works on messaging and email. Refer to Known Limitations of the MCP Server for which AI agent tickets can be listed.
Help Center Work
- Compare the questions customers asked in solved tickets last month with our Help Center articles. Which topics are answered manually again and again with no article?
- Based on ticket volume, which 10 articles should we write next?
- Read the articles in our Billing section, then draft a new article about invoice changes in the same structure and tone, as a draft in that section.
- Update the Spanish translation of article 12345 to match the latest English version.
Refer to Replying to Tickets and Updating Help Center Articles with the MCP Server before making changes to published articles.
Drafting Replies and Working the Queue
- Read ticket 12345 and add a suggested reply as an internal note.
- Add an internal note to ticket 12345 with the invoice the customer sent on ticket 12300 attached.
- In Claude Code: add an internal note to ticket 12345 explaining how to regenerate the API token, with the screenshot api-settings.png from my desktop placed after the first step.
- Create a ticket for jane@example.com about her refund request, with a private note summarizing our call, and assign it to the Billing group.
- Go through the unassigned open tickets, suggest a priority and group for each, and list them for me before changing anything.
- Tag all tickets from the last week that mention the new pricing with the tag pricing_feedback.
- Ticket 12345 is under the wrong organization. Find the organization whose external ID matches the Account ID field on the ticket, add the requester to it if needed, and move the ticket.
- A customer messaged us from +62 812 3456 7890. Find them by the mobile number in their user tags and show me their open tickets.
- Show me the end users tagged vip, with their organization.
- Tickets 12345, 12346 and 12350 report the same login error. Create a problem ticket for it and link the three tickets to it as incidents.
- The login fix is deployed. Solve problem ticket 12400 with an internal note, and tell me first how many incidents it will solve.
- Find tickets from the same customer as ticket 12345 in the last week that look like duplicates of it.
- Merge tickets 12346 and 12350 into 12345, with an internal note on each.
Merging needs Allow merging tickets. Refer to Write Permissions for the MCP Server.
Curating and Applying Macros
- Find every macro that mentions the old refund policy and show me what each one says.
- List our active macros that nobody used in the last 30 days.
- Find macros that say almost the same thing and suggest which ones to keep.
- Rewrite the reply in the "Refund approved" macro to be shorter and friendlier. Show me the change before saving it.
- Deactivate the unused macros we just agreed on.
- Preview the "Password reset" macro on ticket 12345, then apply it as an internal note.
Changing macros needs Allow changes to macros. Refer to Write Permissions for the MCP Server.
Combining Zendesk with Other Tools
If your AI assistant is connected to other tools, it can use them together with the MCP Server:
- Look up a customer's order in your e-commerce connector while drafting a reply in Zendesk.
- Create a bug in Jira from a support ticket, with the full ticket context.
- Combine ticket history with meeting notes and emails to get a complete view of a customer.
What Customers Say
From reviews in the Zendesk Marketplace:
"We've automated 90% of our Help Center processes with this MCP, having workflows to spot gaps, update or create existing articles directly in Zendesk, with minimal approval from us."
"We connected Claude AI to our Zendesk account using this MCP Server and it has completely transformed our customer support workflow."
For tips on getting better results, refer to Using the MCP Server.
If you need help, please contact Swifteq support using the Live Chat or email support@swifteq.com.