Rovo agent
The Quartz SLA agent answers SLA questions in plain language inside Rovo Chat — what an issue's status is, and what is about to breach — using only Quartz's own SLA numbers and only the issues the asker can already see. Advanced
The Rovo agent is part of the Advanced edition. On a Standard install the agent stays visible but inert: its actions refuse with a message until AI is entitled. The agent only appears at all once Rovo is activated on your Atlassian site.
What it is
Quartz ships a Rovo agent named Quartz SLA that lives in Rovo Chat. You ask it about service levels in everyday language and it replies from the same figures the rest of Quartz computes — the live SLA status of a given issue, or the open issues that are at risk of breaching.
The agent answers only from Quartz's own SLA results; it does not read or reason about your ticket content, and it will not invent issues, numbers, dates or causes.
Example prompts
Two conversation starters appear when you open the agent, and you can type your own questions in the same vein:
| Prompt | What it answers |
|---|---|
| What SLAs are about to breach? | Open issues at risk of breaching, soonest first. You can narrow to a window (next 8, 24, or 48 hours). |
| Is this issue going to breach its SLA? | The current SLA status for a single issue you name by key. |
You can phrase the same questions naturally — for example asking about a specific issue key, or asking what is at risk in the next day. The agent maps your wording onto its two actions and replies from their results.
The two actions
Everything the agent reports comes from Quartz's own SLA data, computed on demand.
Status of one issue
quartz-get-sla-status takes a single issue key and returns its applicable SLA goals and their current status — the same status, elapsed, and remaining figures you would see on the issue panel.
What is at risk
quartz-at-risk returns open issues at risk of breaching, sorted by soonest breach, with an optional time window of 8, 24, or 48 hours. This is the same pause-aware view the Radar's "what breaches next" board is built from.
It respects the asker's permissions
Both actions enforce the asking user's own Jira access before revealing anything. The single-issue action checks that the asker can browse that issue; the at-risk action runs its scan as the calling user. Jira itself filters the results to the issues that person is allowed to see, so two people asking the same question can get different lists.
The agent is user-scoped just like the Radar and Reports. Sharing the agent with your team does not expose issues anyone could not already browse in Jira.
On Standard
Because the agent is Advanced-only, both of its actions refuse with a clear message when AI is not entitled. The agent does not disappear on Standard — it stays listed but inert, so the lock is visible rather than silent. In production the Marketplace licence is the sole authority for whether AI is enabled; there is no override that unlocks it without the Advanced edition.
Where it runs
The agent runs on Atlassian's Rovo. Like every AI feature in Quartz, it uses Atlassian-hosted Claude through the Forge runtime, so prompts never leave Atlassian's cloud and no issue data is sent to a third party.
Was this page helpful? Yes · No · Contact support