Request Router
A Rovo agent, formerly named Service Triage Agent, that analyses incoming Jira Service Management requests to suggest request type, priority and escalation, draft replies and summaries. Atlassian documents skills that update a work item's request type and add a summary as a comment. In automation rules, Atlassian's current documentation is internally inconsistent about whether out-of-the-box agents may act or only return text consumed by a separate automation action.
Recorded characteristics
- Function
- Atlassian documents Request Router (formerly Service Triage Agent) as a Rovo agent that automatically triages incoming requests by analysing their content, sentiment and other details to determine request type, urgency and priority, and to determine whether a request should be escalated based on SLAs, customer interactions and urgency. Documented use cases are: generate reply (drafting a tailored reply from details and comments of past similar work items), suggest priority, suggest request type, suggest escalation, generate summary, and generate case summary. Atlassian's Skills table for this agent lists four supported skills: Suggest priority; Suggest request type, described as "Update the request type for a work item based on similar work items in a Jira Service Management space"; Summarize work item, described as "Create a comprehensive summary of a work item in a Jira Service Management space and add it as a comment"; and Suggest escalation. Atlassian states the agent is accessible within Jira, Jira Service Management, Jira Product Discovery and Confluence, but generates higher-quality output in the context of service requests in Jira Service Management, and that it continuously learns from past similar work items.
- Data access
- Atlassian establishes, for this agent specifically, that it reads request content and sentiment; details such as summary and description (used to suggest request type); details, comments and customer sentiment (used to suggest priority); details and comments from past similar work items (used to draft replies); customer sentiment, SLAs and other details (used to suggest escalation); the overall activity and conversations in a work item (used to generate a summary); and associated data, notes and customer communications across multiple interactions and agents (used to generate a case summary). Its skills are scoped to work items in a Jira Service Management space, and Atlassian states it draws on similar work items in that space. Generic Rovo data access beyond these documented sources is not attributed to this agent.
- Actions
- Can take actions
- External actions
- Unknown
- Human confirmation
- Conditional
- Permission basis
- User permissions
- Administrative control
- Documented controls: selecting the agent from the Rovo agent list in Rovo Chat; per-agent skills, which determine which actions the agent can carry out; using the agent in an automation rule via the "Use agent" action, which requires connecting Rovo to automation; adding "Automation mode" at the start of the prompt to use the agent's response within an automation rule; automation trigger configuration on the agent (Triggers screen in Studio, available to an agent's manager or editor) or adding the agent as an action inside an automation flow; an explicit organisation-level ability for admins and users to prevent agents from acting in automations, after which any agent tool performing a write action fails and flows must use the {{agentResponse}} smart value instead; Atlassian's statement that agents used in automations can be configured without requiring user confirmation for every action; Studio admin control over who may create agents (All users, Selected groups of up to 10 groups, or No users); per-agent user access (Open to all users toggle, plus individually added Editors and Managers, with group and team restriction not supported); author display name attribution; and thumbs-up / thumbs-down feedback on the agent's suggestions. Rovo credit usage is tracked and managed at organisation level by organisation and billing admins.
- Default state
- Not established
- Availability
- Atlassian documents Request Router as an out-of-the-box Rovo agent accessible within Jira, Jira Service Management, Jira Product Discovery and Confluence, presented in the Rovo agent browsing directory with three conversation starters. Atlassian's Rovo agent governance documentation states Rovo Studio availability as Atlassian Cloud Standard, Premium and Enterprise, and records Atlassian Government Cloud as not available. Atlassian does not label this capability preview or generally available on the capability page, and no preview or GA status is inferred here.
- Licensing
- Atlassian states that Rovo credits are its unit of measurement for usage of AI capabilities and access to enriched Teamwork Graph context, that credit usage depends on the complexity of a user's or an agent's interactions with AI features, and that organisations receive Rovo credit allowances based on their Atlassian app and plan, pooled at organisation level. Atlassian records that extra usage billing for Rovo credits takes effect on 3 December 2026. No capability-specific licence for Request Router is documented.
- External model or provider
- Not established
- Limitations and uncertainty
- Recorded boundaries and open questions: (1) Atlassian's user-facing descriptions are phrased as suggestion and determination ("suggest priority", "suggest request type", "suggest escalation", "determine whether a request should be escalated"), while the Skills table for the same agent describes updating the request type and adding a summary as a comment; the action classification here rests on the Skills table wording, not on the marketing phrasing. (2) Escalation is documented only as a suggestion or determination; no escalation execution by the agent is documented and none is claimed. (3) Generate reply is documented as drafting a reply that the user may use as is or edit; Atlassian does not document the agent sending or posting the reply to a customer. (4) Atlassian's current documentation is internally inconsistent about automation: the tools page states that an agent triggered in an automation flow cannot use its own tools and can only provide a text response used via the {{agentResponse}} smart value, and the automations page likewise states that out-of-the-box agents do not take actions and that actions must be added after the agent step; yet the same automations page states that using an agent within an automation flow allows the agent to take actions directly in the apps it has access to, and that agents in automations can be configured without requiring confirmation for every action. Request Router is an out-of-the-box agent. This inconsistency is recorded rather than resolved in favour of the more permissive reading. (5) Consequently, whether any Jira field write in an automated triage flow is performed by Request Router itself or by a separate downstream automation action is not settled by current documentation. (6) Human confirmation is recorded as conditional because Atlassian documents interactive confirmation before consequential tools that may mutate data, while agents in automations can be configured without per-action confirmation. (7) Permission basis rests on Atlassian's statement that tools and skills always respect the permissions of the person using the agent; the runtime identity for automation execution, where Rovo is connected to automation "to act on your behalf", is not separately specified for this agent. (8) No default operational state is established: Atlassian documents neither an enabled-by-default automatic triage behaviour nor an explicit disabled state, and automatic triage requires an automation rule to be deliberately configured. (9) Atlassian does not identify the model or provider used by this agent. (10) No third-party or non-Atlassian action is documented for this agent; generic Rovo third-party tools are not attributed to it. (11) Atlassian states the agent produces higher-quality output within Jira Service Management, and recommends regularly reviewing its suggestions. (12) Rovo credit consumption and its December 2026 extra usage billing may constrain operation.
Evidence
- Work with Request Router
Supports: Function · Data access · Actions · Availability · Limitations · Primary source
Describes the agent as automatically triaging incoming requests by analysing content, sentiment and other details to determine request type, urgency and priority, and to determine whether a request should be escalated; lists generate reply, suggest priority, suggest request type, suggest escalation, generate summary and generate case summary; records the former name Service Triage Agent.
Establishes the inputs: request content and sentiment; summary and description; details, comments and customer sentiment; details and comments from past similar work items; SLAs and customer interactions; overall activity and conversations in a work item; associated data, notes and customer communications.
Skills table states the Suggest request type skill will "Update the request type for a work item based on similar work items in a Jira Service Management space" and the Summarize work item skill will "Create a comprehensive summary of a work item in a Jira Service Management space and add it as a comment" — persistent Jira writes attributed to this agent. Suggest escalation is documented only as suggesting whether a work item should be escalated. Generate reply is documented as drafting a reply the user can use as is or edit.
States the agent is accessible within Jira, Jira Service Management, Jira Product Discovery and Confluence, with higher-quality output for service requests in Jira Service Management; presented among out-of-the-box Rovo agents with three conversation starters.
Recommends regularly reviewing the agent's suggestions and giving thumbs-up or thumbs-down feedback; notes prompts perform better when the work item key or URL is specified; phrases outputs as suggestions while the Skills table describes updates and comments.
- Automate Rovo agents
Supports: Actions · Human confirmation · Admin controls · Default state · Limitations · Primary source
States that out-of-the-box agents do not take actions and that actions must be added to the automation after the agent step, while also stating that using an agent within an automation flow allows the agent to take actions directly in apps it has access to — the inconsistency recorded under limitations.
States that when agents are used in automations they can be configured without requiring user confirmation for every action.
Documents adding an automation trigger in the agent's configuration (manager or editor role), adding the agent to a flow via the Use agent action with a required prompt, connecting Rovo to automation, and the ability of admins and users to prevent agents from acting in automations — after which any write-performing tool fails and flows must use {{agentResponse}}.
Automatic operation requires a deliberately configured automation trigger or flow; no default automatic triage behaviour is documented.
Establishes the {{agentResponse}} pattern in which the agent returns text consumed by subsequent automation actions, leaving the writer of any field update in an automated flow unsettled.
- Add tools to Rovo agents
Supports: Actions · Human confirmation · Permission basis · External actions · Limitations · Primary source
States that if an agent is triggered in an automation flow it cannot use its own tools and can only provide a text response used in subsequent automation actions via the {{agentResponse}} smart value.
States the agent will respond asking for confirmation before executing consequential tools that may mutate data across systems.
States that tools and skills always respect the permissions of the person using the agent, and that the agent can only access information and perform actions the user already has permission to do.
Documents third-party tools (for example Gmail and Google Calendar) as an opt-in connection made by the individual user for their own agents, appearing as performed by the human user; no third-party tool is documented for Request Router.
Records tool limitations such as bulk Jira actions limited to 20 work items and required-field prompts, and the general dependence of agent actions on the acting user's permissions.
- Rovo agent permissions and governance
Supports: Permission basis · Admin controls · Availability · Primary source
States that when someone interacts with an agent, the agent is acting on that person's behalf and can only return or interact with information the user has permission to access — if the user cannot comment, the agent cannot comment.
Documents Studio admin control over who can create agents (All users, Selected groups, No users), Editor and Manager roles, the Open to all users access toggle, and author display name attribution.
Records Rovo Studio agent governance as available on Atlassian Cloud Standard, Premium and Enterprise, and not available on Atlassian Government Cloud.
- How Rovo credits work
Supports: Licensing · Primary source
States that Rovo credits measure usage of AI capabilities and Teamwork Graph context, that usage depends on the complexity of a user's or agent's interactions, that allowances are based on the Atlassian app and plan and pooled at organisation level, and that extra usage billing takes effect on 3 December 2026.