One triage console for every Jira service desk queue

Build me a triage console my support team can live in all day. It should merge the queues of every Jira Service Management service desk we run (IT, HR, Facilities and any others) into a single list ranked by how much SLA time is left, instead of making agents hop between project queues.

The main screen is the board. A handler builds it by calling List Service Desks to find every desk the connected account can see, Get Service Desk Queues for each desk, and Get Issues in Queue to pull the tickets sitting in those queues, falling back to List Customer Requests where a desk has no useful queue defined. For each ticket the handler calls Get Request SLA Information to read the live countdown on the time-to-first-response and time-to-resolution goals, and uses the smallest remaining time as the ticket's sort key. Rows show the ticket key, summary, service desk, request type, current status, reporter, assignee and the remaining time as a human countdown like 43m left or Breached 2h ago.

Ranking and colour rules: sort ascending by time remaining so the most urgent ticket is always the top row. Colour code each row by time remaining, red when under one hour, amber when under four hours, neutral above that. Pin unassigned tickets and already breached tickets to the top of the board above the normal ordering, since those are the ones that quietly rot. Give the board filters for service desk, request type and status, with request types loaded from Get Service Desk Request Types so the filter shows real names rather than free text.

Persist a per-user preference for which service desks that person covers, and open the board pre-filtered to their own slice every time they load it, with a clear control to widen back out to all desks. Two agents on different desks should get different boards from the same app.

Clicking a row opens a detail pane on the right. It loads the ticket with Get Customer Request and the whole conversation with Get Request Comments, showing public replies and internal notes distinctly and in chronological order, so the full history is on screen without opening Jira. The pane has a reply box with a toggle between Reply to customer and Internal note, and posting either one calls Create Request Comment with the public flag set accordingly. It also has a status control: load the available next steps with Get Request Transitions and move the ticket with Perform Request Transition, then refresh the row in the board so the change is visible immediately.

Add a Draft the reply button in the detail pane that kicks off a background agent for that one ticket. The agent reads the request with Get Customer Request and its full thread with Get Request Comments, works out what the customer is actually asking, searches our help centre with Search Knowledge Base Articles for relevant material, and writes a suggested reply plus the matching article link back into the app. The draft lands in the reply box for the agent to edit and send. It must never post to the customer by itself, a human always presses send. Show a clear working state on the button while the draft is being generated, and show the suggested article as a link next to the draft so the agent can check it before citing it.

Add an Escalate to engineering button in the detail pane. It opens a dialog for the target project, issue type and a short description prefilled from the ticket summary and thread, then creates a Jira issue with Create Issue, links that new issue back to the service desk ticket with Create Issue Link using a Relates or Blocks relationship, and posts the escalation to our support Slack channel with Send a Message including the service desk ticket key, the new engineering issue key, the customer impact and a link back to this app. Record the escalation on the ticket as an internal note with Create Request Comment so the history stays in one place.

Implementation notes. Every Jira and Jira Service Management call needs a cloudId, resolved once from the accessible resources endpoint and reused. List responses page with start, limit and isLastPage, so page through queues and requests rather than assuming a single page. Comments can be public or internal, so the reply box has to send that flag explicitly and the thread has to render the difference obviously, since posting an internal note publicly is the worst mistake this app could make. SLA lookups are per ticket, so fetch them concurrently with sensible batching and cache briefly to keep the board fast on a busy morning. Nothing in this app runs on a schedule, this is a working surface a person opens, and the drafting agent only runs when someone presses the button.