Release notes
Release notes for v1.3.0
New features, improvements, and fixes in alterac.ai v1.3.0.
Your whole team's work now lives in one place. v1.3.0 introduces Person tasks, so the work people do themselves sits right beside the work your coding agents deliver, in the same projects, Plans, and review flow. It also adds a Closed state for work you decide not to finish, and gives coding agents a clearer start on every task.
Not every task belongs to a coding agent. Now you can create a Person task for work a teammate does themselves, like a manual check, a design decision, or a customer follow-up.
- Choose Coding agent or Person when you create a task.
- Select Start work to take a Person task and track it In progress.
- Submit your work for review with a summary, then accept it, send it back for more work, or complete it directly.
- See Person tasks alongside agent work in project home, Tasks, Plans, and metrics. The Working count covers both people in progress and agents at work.
- Follow one history for each task that shows every attempt, whether a person or a coding agent did the work.
Person tasks use Weight to describe the expected difficulty of the work itself. Coding agent tasks keep Mass, the expected difficulty of reviewing what the agent produced. The Tasks list shows both in one Mass/Weight column, and the CLI accepts --weight for Person tasks.
Priorities change. Close a task to take it out of your way without deleting its history, and reopen it any time.
- Closed tasks leave the normal task view, so your list stays focused on active work.
- Turn on Show closed when you want to see them again.
- Closed tasks do not count against your completion metrics or Plan progress.
When Task Builder breaks down a larger piece of work, choose whether each draft is a Coding agent or Person task. Each draft shows its task-type icon, so you can review who does what before you publish.
Coding agents now receive a short alterac.ai launch message when they start a task. It points them straight to your repository's alterac.ai process and their exact work run, so they begin in the right place every time.
- Write multi-line coding-agent commands using backslash line continuations.
- Leave review comments of any length.
- Reopen a task with one click; it saves immediately.
- See human review time as N/A when working hours are not configured, and a friendlier empty state on a new project's metrics.
- Fixed duplicate Plan labels in Task Builder review.
- Fixed review history for tasks whose coding-agent work started before this release.
Update each coding-agent command to use {alterac-ai-message} in place of {task-work-run-id}. A command that still uses the old placeholder cannot queue new work, and work queued before this release needs to be requeued after you update the command. See Write a coding-agent command.
Install the latest CLI with this release. Older CLI versions can clear a task's Mass or Weight when they update the task.
Learn more in Understand the task workflow and Organize tasks.
