This article lists the tools the Swifteq MCP Server gives your AI assistant, and the limits that apply. Your AI assistant chooses the right tool from your request, so you do not need to name tools, although naming them can help in ChatGPT.
In this Article
Ticket Tools
Organization and User Tools
Macro Tools
Help Center Tools
CSAT
Agent and Configuration Tools
Account Tools
Tools that Make Changes
Limits
Ticket Tools
| Tool | What it does |
|---|---|
| get_ticket | Gets one ticket with its fields, including status, priority, requester, assignee, group, organization, brand, tags, custom fields, satisfaction rating, the people CC'd on it and a link to the ticket. Can also return the ticket's organization, with its external ID and custom fields, the requester's details, for a problem ticket the incidents linked to it, and the ticket's SLA targets: whether each was achieved or breached, is still running or paused, and when the next one breaches. It can also say whether a Zendesk AI agent took part: whether it handled the ticket alone or handed it over to a person, when and after which message, and whether Zendesk counted the ticket as an automated resolution. |
| search_tickets | Searches tickets by text, created or solved date, status, custom status, priority, ticket type, tags, attachments, brand, group, organization, assignee, requester, ticket form or a custom field value. Can also find Zendesk AI agent tickets on messaging: the ones the AI agent resolved alone, the ones it escalated to a person, and the ones Zendesk counted as automated resolutions. Counting tickets takes this filter too; metrics and SLA reports do not yet. Returns only the ID, status and created date unless you ask for more fields. Can also return each ticket's organization and requester, its metrics such as first reply and resolution times, and its SLA targets. Can return results as CSV. |
| count_tickets | Counts tickets using the same filters as search. Much faster when you only need a number. Can also split the count in one request by status or custom status, priority, ticket type, brand, group, form, channel, a drop-down, multi-select or checkbox field, CSAT rating, or day, week or month. Split by CSAT rating, it also gives your CSAT score and how many surveys went unanswered. |
| get_ticket_metrics_summary | Reports ticket metrics for a period in one request: first reply time, first and full resolution time, requester wait time, agent wait time and on-hold time (median, average and 90th percentile, in calendar and business hours), reopen rate, replies per ticket and one-touch rate. Uses the same filters as search, and can split the figures by group, brand, agent, channel, status or custom status, priority, ticket type, form, a drop-down, multi-select or checkbox field, CSAT rating, week or month, or add a day-by-day trend. |
| get_sla_queue | Lists the unsolved tickets that most need attention under your Zendesk SLA policies: first the tickets with a target overdue now, then the rest by their next breach. Reads the whole backlog, not one page, and uses the same filters as search, for example your own tickets or one group. |
| get_sla_summary | Reports SLA achievement for a period in one request: for each SLA metric (first reply, next reply, periodic and pausable update, requester wait, agent work and total resolution time), the achieved and breached targets and the percentage achieved. Each target counts once, so a ticket with three customer follow-ups has three next reply targets. Uses the same filters as search, can split the figures by group, agent, brand, channel, priority, SLA policy, week or month, and lists the breached targets. |
| get_ticket_conversation | Gets the full conversation of a ticket, including every message, who wrote it (the customer, an agent, a Zendesk AI agent or Zendesk itself), whether it is public or internal, and its attachments. When a Zendesk AI agent took part, it also shows whether a person took over and at which message. Works for email, web form, messaging, WhatsApp and other channels. |
| get_ticket_attachment | Downloads one attachment. Images are returned so AI assistants that support images can read them. |
| create_ticket | Creates a new ticket. The first comment is a private internal note unless you ask for a public one. The ticket can be for someone already in Zendesk, for a new person from their email address, or an internal ticket from you. Sets the same fields as update_ticket, plus brand and ticket form. Can also CC people on a ticket with a public first comment, attach files to the first comment and place images in it. Can also create the ticket as an incident linked to a problem ticket. |
| update_ticket | Changes status, custom status, type, priority, subject, assignee, group or organization, adds a comment with files attached or images placed in it, adds or removes tags, adds or removes CCs, updates custom fields, and links a ticket to a problem ticket as an incident or unlinks it. |
| merge_tickets | Merges up to 5 duplicate tickets into one ticket, as an agent does in Zendesk. Each duplicate is closed, with a comment on it and on the ticket kept, internal unless you ask for public replies. Only tickets from the same requester. Cannot be undone. Needs Allow merging tickets. |
| request_ticket_attachment_upload | Gives your AI assistant a short-lived link to upload a file from your computer, to attach to a comment. Only needed for files on your computer: files already in Zendesk are attached directly. |
If your account uses custom ticket statuses, ask your AI assistant to list the ticket fields first. It can then filter and update tickets by the exact status name.
Organization and User Tools
| Tool | What it does |
|---|---|
| search_organizations | Finds organizations and returns their details, including external ID, domains, tags, notes and custom fields. Looks them up by ID, by external ID (for example an account ID from your CRM), by exact name, by the start of the name or by tag (the whole tag, or its start or end), or lists them all. |
| search_users | Finds users, customers and agents, and returns their details, including email, role, external ID, organization and custom fields. Looks them up by ID, external ID, email, name, role or tag (for example the end users tagged vip, or a customer whose mobile number is stored as a tag), or lists the members of an organization. Only a user's own tags count, not their organization's. Can also show every organization each user belongs to. |
| add_user_to_organization | Adds a user to an organization, for example so a ticket can be moved to that organization. Zendesk only lets you set a ticket's organization to one its requester belongs to. Needs Allow changes to users and organizations, which is off by default. |
To see the organization and requester of many tickets at once, ask for them in the same ticket search, for example list this week's open tickets with each organization's ARR. The AI assistant then gets them in one request instead of one per ticket.
Macro Tools
| Tool | What it does |
|---|---|
| search_macros | Finds and reads macros: by words in the title, the reply text or the dynamic content translations a macro uses (each result says where it matched), by ID, or lists them all. Filters by active or inactive, who can use the macro (all agents, a group or one agent) and category. Sorts by name, date, position or usage in the last hour, day, 7 or 30 days, which finds unused macros. Returns each macro's actions in plain words, and the reply text when you ask for it. |
| preview_macro | Shows what applying a macro would do to a ticket, without changing anything: the fields it would change, the reply as the customer would get it with placeholders and dynamic content filled in, and the files it attaches. |
| apply_macro | Applies a macro to a ticket as an agent does in Zendesk: the reply with placeholders filled in, the field changes and the macro's files. The reply is an internal note unless the macro makes it public or you ask for a public reply, which needs Allow public ticket replies. Applying the same macro to the same ticket again within 10 minutes sends nothing unless you ask. The macro is recorded in Zendesk, so its usage counts include it. |
| create_macro | Creates a macro for every agent, for a group, or only for the Zendesk user behind the connection. Needs Allow changes to macros. |
| update_macro | Changes a macro's wording, actions, title, who can use it or its position. Shows each change with its old and new value, so it can be changed back. Needs Allow changes to macros. |
| set_macros_active | Deactivates or reactivates up to 100 macros at once. Deactivating is how a macro is removed: agents no longer see it, and it can be reactivated at any time. Needs Allow changes to macros. |
A change to a shared macro changes what every agent sends from that moment. Macros cannot be deleted through the MCP Server. Refer to Replying to Tickets and Updating Help Center Articles with the MCP Server.
Help Center Tools
| Tool | What it does |
|---|---|
| search_help_center_articles | Searches articles in one or more languages. Returns titles, links and snippets, not the full content. |
| get_help_center_article_content | Gets one article with its full content. |
| export_help_center_articles | Exports many articles with their full content, filtered by language, category or section. |
| create_article | Creates an article in a section, as a draft by default. |
| update_article_translation | Updates the title or content of an article in any language, and creates the translation if it does not exist. Saved as a draft by default. |
| list_help_center_categories and list_help_center_sections | List categories and sections. |
| create_help_center_category and update_help_center_category | Create a category, or change its name, description or position. |
| create_help_center_section and update_help_center_section | Create a section, or change its name, description, position or category. |
| request_help_center_image_upload and create_help_center_image | Upload an image, such as a screenshot, to use in articles. |
| list_permission_groups and list_user_segments | List who can edit articles and who can view them, for use when creating articles. |
If you have more than one brand, the Help Center tools can work with any brand's Help Center. Ask your AI assistant to list your brands first.
CSAT
| Tool | What it does |
|---|---|
| get_satisfaction_ratings | Gets CSAT ratings filtered by date, score or ticket. Works with both Zendesk's original satisfaction ratings and the newer customizable surveys, and detects which one your account uses. For a CSAT score over a period, count_tickets split by CSAT rating gives the score and the counts in one request. |
Agent and Configuration Tools
| Tool | What it does |
|---|---|
| find_agent_by_name | Finds an agent by name, for example to assign a ticket. To find customers and other users, the AI assistant uses search_users. |
| list_brands | Lists your brands and their Help Centers. |
| list_ticket_fields | Lists system and custom ticket fields, including drop-down options and custom statuses. |
| list_ticket_forms | Lists ticket forms and their fields. |
| list_triggers | Lists triggers with their conditions, actions and usage, which is useful for reviewing automations. |
Account Tools
| Tool | What it does |
|---|---|
| get_usage | Reports your MCP Server usage this month: calls made, free allowance, calls remaining, when it resets and your plan. Using this tool does not count against your allowance. |
| submit_mcp_feedback | Sends feedback about the MCP Server to Swifteq, for example a capability you needed that is missing, a result that was wrong or confusing, or a task that took too many steps. Your AI assistant shows you what it will send and sends it only after you agree. Using this tool does not count against your allowance. Refer to Using the MCP Server. |
Tools that Make Changes
The following tools make changes in Zendesk: create_ticket, update_ticket, merge_tickets, request_ticket_attachment_upload, create_article, update_article_translation, create_help_center_category, update_help_center_category, create_help_center_section, update_help_center_section, request_help_center_image_upload, create_help_center_image, add_user_to_organization, apply_macro, create_macro, update_macro and set_macros_active. All other tools only read. submit_mcp_feedback does not touch Zendesk: it sends your feedback to Swifteq.
The tools that make changes need Write Helpdesk Data access, and they are all refused when Allow any changes in Zendesk is off. Public replies, article publishing, changes to users and organizations, changes to macros and merging tickets have their own permissions. Refer to Write Permissions for the MCP Server.
All tools are listed in your AI assistant even when you only have read access. Tools that make changes fail if they are used without write access.
Limits
| Limit | Value |
|---|---|
| Results per page | 100 (your AI assistant can request more pages) |
| Rows per CSV export | 1,000 |
| Tickets per metrics summary | 3,000 (split longer periods, for example into weeks) |
| Days per metrics trend | 31 |
| Tickets per SLA summary or SLA queue | 2,000 (split longer periods, for example into weeks, or narrow the queue by group) |
| Metrics summaries, SLA summaries and queues, and CSV exports running at the same time | 2 per Swifteq account |
| Values per split count | 200 (for example 200 groups or 200 days) |
| Comment length | 10,000 characters |
| Search text length | 500 characters |
| Ticket subject length | 500 characters |
| CCs per request | 20 |
| Article title length | 500 characters |
| Article content length | 100,000 characters |
| Attachment and image size | 20 MB |
| Files attached to a comment | Up to 10 per comment, each up to 20 MB or your Zendesk account's attachment limit if lower |
| Macros searched per request | 1,000 (your AI assistant can continue the search) |
| Macros activated or deactivated per request | 100 |
| Tickets merged per request | 5, into one ticket |
| Feedback sent to Swifteq | 10 per person per day |
| Requests per minute | 1,200 |
If Zendesk asks the MCP Server to slow down for a few seconds, it waits and retries once. For a longer wait, your AI assistant is told how many seconds to wait before trying again.
If your account already has two metrics summaries, SLA summaries or queues, or CSV exports running, the next one is not started and your AI assistant is told to try again in a few seconds. Ask for these one after another rather than all at the same time.
If you need help, please contact Swifteq support using the Live Chat or email support@swifteq.com.