Projects give agents a plan,
context, and a finish line.
Move from one-off chat runs to persistent project work. Create strategies, tickets, releases, reviews, resources, and agent threads that stay connected until the outcome is shipped.

Not a chat window.
A working agent.
A project gives every agent the surrounding plan: strategy, tickets, releases, files, comments, reviewers, resources, and run history. Agents can return to the same workspace and continue the work instead of restarting from a prompt.


















































Not a chat window.
A working agent.
Projects keep the plan, tickets, files, reviews, resources, and run history connected so agents can continue from the actual state of the work.

















































Turn a goal into a strategy, release, and backlog.
Mission Control analyzes the project, creates a strategy, breaks the work into tasks, groups them into releases, and can return later to clean up scope as the project changes.
Manage agent work like real product work.
Tasks move through backlog, board, review, and finished states. Every ticket can carry detail, assignees, reviewers, blocked-by relationships, schedules, environments, comments, and related threads.
Let humans or agents review work before it ships.
A task can require review by a human or an agent. Review agents can start their own thread, inspect the project context, request changes, or approve the work into a finished state.
Start tasks one after another until the project is done.
Projects are designed for more autonomous execution. Agents can work through tasks, decide when subtasks need separate threads, trigger Mission Control, and continue from the project state they leave behind.
Give every agent the project context it needs to finish.
Projects connect strategy, backlog, releases, computers, files, reviews, and threads so complex work can keep moving after the first message.
Get started