Backlog and board hold the same Tasks
The Backlog holds work that is not committed yet; the board shows committed work moving. Same Tasks, two views, one source.
HiFoxAgile development
Work in the Sprintdelivered by people and Agents
The team runs its Sprints the way it always has. Once scope is committed, Agents sit alongside the other assignees, sharing the same capacity and the same burnup. Acceptance criteria live in the Task description, so progress and acceptance stay inside the Sprint.
How it works
Agents belong in the Sprint
Backlog, boards, and acceptance criteria are practices the team already has. Agents take the execution slot. Work without acceptance criteria does not enter the Sprint.
Agents sit next to the teammates executing
A Sprint is a Space time box. Agents and teammates share one Assignee control, and Agents count toward capacity too.
Acceptance criteria are written into the Task
The description spells out what done means, and the Sprint closes against it. Until that is written down, the Task is not ready.
Agile Workflow
Agent output lands back in the Sprint
Scope, progress, execution, and acceptance all write to Tasks. There is no second board reserved for Agents.
01Task board
The board lays out the cycle's work
Committed Tasks sit split by status on one board, so who is on what and what is still open is plain at a glance.
- The board holds this cycle's committed work
- Split by status, so each column reads at a glance
- Regroup by Sprint, assignee, or Project
02Sprints
Scope, capacity, and burnup in one place
A Sprint is a Space time box. Once it starts, work in progress, scope changes, and capacity stay visible, and burnup updates as Tasks move.
- Scope, started, completed, and capacity on one screen
- Burnup updates as Tasks move, with no separate report
- Remaining scope and blockers are visible at the end of the cycle
03Projects
Follow cross-Sprint goals as Projects
A Project groups related Tasks for one goal or release. The Tasks themselves stay in their own Space and Sprint.
- A Project can span several Sprints
- With estimates on, the Project rolls up estimate points
- Filter boards by Project without duplicating Tasks
04Agents executing work
Put an Agent in the Sprint's execution slot
An Agent picks up committed Tasks from the Assignee control, consuming Sprint scope and counting toward burnup just like a teammate. Execution runs in its own worktree, and progress, blockers, and results return to the Task.
- Assigning an Agent and a teammate is the same control
- Agent work counts toward scope, capacity, and burnup
- Execution stays in its own worktree, off the main workspace
- Unfinished Agent work rolls forward with everything else
05Automated testing
Write the criteria into the Task, and the Agent runs them
The criteria in the description are the test scope. Agents run regression inside the Sprint and return results and evidence to the Task, where a failing case becomes work for the same cycle.
- Criteria live in the description, and tests run against them
- Failing cases return as Tasks, with no ticket to open by hand
- Run mode and environment are recorded with the results
- Cases added this cycle are next cycle's regression scope
Free to use
Use your existing Agent tools and Coding Plan
Directly connect Claude Code, Codex, OpenCode, OpenClaw, Hermes, Pi, and other Agent tools that support CLI usage. Token usage goes through the Coding Plan or API key you already have in your Agent tools, and APIs from model providers can also be connected and used directly.
The agile system
The concepts an agile team already knows
Backlog, Sprint, Project, and View define how a Space runs agile. An Agent takes the execution slot on a Task; it is not an extra layer.
Tasks
Goal, type, status, priority, estimates, assignees, and comments stay on one Task.
Backlog
Uncommitted work: ordered, typed, and waiting for the next Sprint.
Sprints
A time box configured per Space, with capacity from recent velocity and burnup history.
Projects
An optional grouping for one goal or release, spanning several Sprints.
Views
Saved boards and filters, so planning and review open the same one.
Testing and acceptance
Criteria live on the Task. Agents run inside the Sprint and return results and evidence to it.
Quality of life, built in
The details that let a Sprint close cleanly
Discussion, execution, and acceptance all attach to the Task.
Blockers marked on board cards
Inbox for Tasks waiting on a reply or review
Comments that can queue the next Agent run
Estimates rolled into Sprint capacity and burnup
Agent test results kept on the same Task
Saved Views for Sprint, review, and personal work
Two-way sync for TAPD and Jira, with GitHub and GitLab changes linked to Tasks
Put an Agent in your next Sprint
Shape the Backlog, commit the scope, and assign what needs executing to an Agent or a teammate.
Start
Stand up an agile Space in three steps
Fill the Backlog, commit a Sprint, and put criteria and an assignee on it.
01
Build the Backlog
Create Tasks with a type, an estimate, and an assignee. Keep what is not ready in the Backlog.
02
Commit a Sprint
Pull in what the cycle can finish. Agents take their Tasks from the same committed set as everyone else.
03
Write the criteria, then assign
Spell out what done means in the description, then assign a teammate to review and an Agent to execute or run regression.
FAQ
Sprints as usual, with Agents on the same board
Does an agile team have to use Sprints?
+
No. Sprints are configured per Space. Use Sprints for time-boxed delivery, and board Views grouped by status for continuous flow. A Space can run both.
How do Sprints work?
+
Cadence is configured per Space. Automatic mode maintains the cycle length, cooldown, and start day; manual mode lets you create, start, and complete one at any time. Unfinished work rolls into the next Sprint.
How do Agents take part in a Sprint?
+
Agents and teammates share one Assignee control. An Agent takes committed Tasks, consuming Sprint scope and counting toward capacity and burnup. Execution runs in its own worktree and returns results to the Task.
How does regression testing run inside a Sprint?
+
Acceptance criteria are written into the Task description, and that is the test scope. Agents run regression against it inside the Sprint, and failing cases become Tasks for the same cycle. Cases added this cycle are next cycle's regression scope.
What is the difference between a Sprint, a Project, and a View?
+
A Sprint is a time box inside a Space, a Project groups related work across Sprints, and a View is a saved way to look at Tasks. A Task can belong to one Sprint and one Project and appear in many Views.
Does adding a Task to a Sprint start the Agent?
+
No. Committing a Task does not change its status. Execution starts when a Task with a ready Agent or Crew leaves the Backlog for a non-terminal status, and the Computer and Runtime are available.
Can we import TAPD or Jira sprints?
+
Yes. Jira import brings in issues, epics, statuses, users, and Sprint data, and TAPD and similar platforms sync two ways, so existing work keeps moving.