Skip to content

Cycles

Project → Cycles. A cycle is a start date, an end date and the tasks that are in it. Sprint, iteration, week — the word does not matter; the line around a promise does.

New cycle, a name, two dates. Then put work in it: drag tasks in from the backlog, or filter the task list to cycle = none and bulk-assign a selection.

One cycle per project is active at a time — the one whose dates contain today. That is what the project header shows, what the burn-up chart draws, and what a new task created from inside the project joins by default.

Progress How many of the tasks in it are done, and how many points
Burn-up Two lines over the cycle: everything in it, and everything finished. They meet when it is done
A filter everything understands cycle = "Sprint 14", on any layout, in any saved view
A line in the past What was in the cycle, and what actually got finished in it

The burn-up is a burn-up, not a burn-down, and the difference is the point: the top line is scope. Scope climbing mid-cycle is work being added, which is worth seeing rather than hiding under a falling line.

Nothing automatic. Kolibri does not roll unfinished work forward, close the cycle or nag you, because every team’s answer to what happens to the three things we did not finish is different and a tool that picks one is a tool people fight.

What most teams do:

  1. Filter the closing cycle to is: open.
  2. Look at each one and decide: next cycle, back to the backlog, or it was never real.
  3. Select the ones going forward and bulk-set the new cycle.

Three minutes, and the record of what the old cycle actually contained stays true.

A cycle normally belongs to one project. Which projects run it in the new-cycle form offers two other answers:

  • Chosen projects — tick the ones that are in it. Web and Mobile share a fortnight; Platform, on its own schedule, never sees it.
  • Every project — the whole workspace, including projects made later.

Either way it is one cycle. That is the point: for a fortnight three teams genuinely share — one planning meeting, one end date — three separate cycles with the same name drift apart within a month, and each project then draws a burn-up of a thing nobody planned per project.

Just this project Chosen projects Every project
Appears in Its project’s Cycles tab The Cycles tab of each one, chipped 2 projects Every project’s Cycles tab, marked Shared
Work in it That project’s Any of theirs Any project’s
Progress shown Its own The whole cycle’s, on every tab — the same cycle showing different numbers depending on where you opened it would be worse The whole cycle’s
Deleting it Affects one project Takes it out of all of them Takes it out of all of them

The scope is fixed when you create it, in the interface. There is no dropdown to re-scope a running cycle, because narrowing one silently strands the dropped projects’ tasks — a change to your data wearing the clothes of a setting. Make the new one and move the work.

An assistant can re-scope one over MCP, and it will tell you what it did: update_cycle returns the tasks left in a cycle that no longer covers their project, by name, having moved none of them.

Cycles and modules are different questions

Section titled “Cycles and modules are different questions”

A cycle answers when. A module answers what is this part of. A task can be in both, and usually is: Sprint 14 and Payments v2.

If you find yourself making a cycle called Payments v2 and stretching its dates whenever the work slips, that is a module trying to get out.

list_cycles, create_cycle, update_cycle and delete_cycle are tools an assistant can use. Combined with project_status and the sprint_planning prompt, “what did not get finished last cycle and what is blocking it” is a question you can ask rather than a report you assemble. See connecting an assistant.