Recurring Work for Jira
Work items that come back on a calendar: daily, on weekdays, weekly on chosen days, monthly on a day or on the nth weekday, or yearly, at a time in a named timezone. Never created twice, never a backlog after an outage.
Getting started
- Install Recurring Work for Jira from the Atlassian Marketplace. It appears under Apps → Recurring work and in every company-managed project's settings under Apps → Recurring work.
- Create a schedule: name, project, what to create (fields you write, or a copy of an existing work item), how often, at what time and in which timezone, a start date and optionally an end date or a maximum number of occurrences. The editor shows the next five dates as you type.
- Let it run. The app checks once every hour and creates what has fallen due. A 09:15 schedule is created between 09:15 and 10:15 local time.
What gets created
With fields you write: the work item type, a summary, a description, a priority, labels and components. With a copy of an existing work item: its type, description, priority, labels and components are read at each run, so editing the original changes future copies; subtasks and attachments are not copied. Every created work item carries a tracking label recurring-<id>, which the history and the skip option rely on.
Variables
In summaries and descriptions: {{date}} (2026-11-09), {{day}}, {{month}}, {{monthName}}, {{year}}, {{week}} (ISO week), {{isoYear}} (the ISO week-based year; 1 January 2027 is week 53 of 2026, so pair {{week}} with {{isoYear}} rather than {{year}}), {{weekday}}, {{time}}. A fallback is written {{name|text}}. Dates are taken from the occurrence in the schedule's timezone, not from the moment the app happened to run.
Calendars
| Repeats | Rules |
|---|---|
| Daily, every N days | Counted from the start date. |
| Weekdays | Monday to Friday; public holidays are not known to the app. |
| Weekly on chosen days, every N weeks | Weeks start on Monday and are counted from the start date's week. |
| Monthly on a day, every N months | Day 31 means the last day of shorter months (30 April, 28 or 29 February); 30 falls on 28 or 29 February. Intervals count from the first month that can fire on or after the start date. |
| Monthly on the nth weekday | First to fourth, or last, of a weekday. |
| Yearly | 29 February falls on 28 February in common years. |
Daylight-saving changes keep the local time: a 09:00 Stockholm schedule stays at 09:00 Stockholm all year, so the UTC moment moves by an hour twice a year.
Options
- Skip while an earlier one is open. If a work item from an earlier occurrence (found by the tracking label) is not yet in a Done status, the occurrence is recorded as skipped and nothing is created. The calendar still advances.
- Assign in turn. Pick people and each occurrence goes to the next one on the list.
- Due date. A number of days after the occurrence date; 0 means the same day.
- Pause and resume. A paused schedule is not created; on resume the next occurrence is computed from now, nothing is back-filled.
- Create now. Creates one occurrence immediately as you, without touching the calendar; it shows in the history as done by hand and does not count toward a maximum number of occurrences.
After an outage
If Atlassian or the app was unavailable and several occurrences were missed, only the latest one is created when the app is back; the earlier ones are recorded in the history as missed. You never get a pile of stale work items. Occurrences more than a day old are also protected by a per-occurrence claim, so two overlapping runs cannot create the same occurrence twice.
If Jira is unavailable
When Jira answers with a temporary error at creation time, the occurrence stays due and the app tries again on the next hourly run; the schedule shows the reason until it succeeds. When Jira refuses the content (for example a required field the schedule does not fill), the occurrence is recorded as failed with Jira's message and the calendar moves on. Fields that are not on the create screen for the chosen type are refused when you save the schedule, not at run time.
Who may do what
| Action | Who |
|---|---|
| See a project's schedules and history | Anyone who can browse the project |
| Create, edit, pause, delete, create now | Project admins of that project, and Jira admins |
| Run the hourly check by hand | Jira admins |
Scheduled work items are created by the app itself, with the app's permissions in the project, so the reporter shows as the app. Manual "Create now" creates as you.
Limits
- 200 schedules per project; 50 people in a round-robin list; 20 labels; one occurrence per schedule per hour.
- History is kept for 400 days.
- Subtasks cannot be created on a schedule; create the parent and add subtasks with Templates for Jira or automation.
Where the data lives
Schedules and history live in Forge storage inside your Atlassian site's region. Nothing leaves Atlassian. See Privacy.
Changelog
- 1.0 (November 2026): first release.
