Mission Control

Mission Control is where you go to understand a single AI employee in depth: what they are thinking about, the work they have done, and the rules and files that govern how they behave. Pick an employee from the selector in the top right, and everything on the page reflects that one person.

Brain state and counters

Across the top you get a quick read on the selected employee:

  • Context items are the directives, notes, and tasks the employee is holding in active memory right now.
  • Active and stale tasks. A task becomes stale when nobody has touched it for a few days, so it surfaces here instead of quietly sitting on the board.
  • Pending notes, proactive queue, and overdue tasks round out the picture.

The brain state panel shows the rules and directives the employee reloads on every reasoning cycle, so you can see exactly which of your instructions fired and when.

Notes, directives, and feedback

You are not stuck with the behavior an AI employee ships with. Use Add to give them guidance:

  • A directive is a hard rule, for example "do not move tasks from Suggested to To Do" or "do not create new proactive tasks while two are already pending."
  • A note or question captures something you want them to keep in mind.
  • Feedback is a softer steer on attitude or style rather than a strict rule.

Remember this is a reasoning system, not a fixed automation. Employees weigh your rules against each other, so write directives clearly and watch how they play out in the brain state.

Activity log

The activity log is a full breadcrumb trail of everything the employee has done: heartbeats, workspace edits, task updates, and more. When you sent someone off on a longer job and want to know what actually happened, this is where you look.

Proactive queue

When an employee is proactive, they generate ideas that contribute to your milestones even before those ideas become tasks. The proactive queue shows what they are considering but have not put on the board, either because they have hit their quota of suggested tasks or have not judged an idea important enough yet. If you spot a good one, ask them to turn it into a task.

Scheduled deliveries

This is where recurring jobs live, like a morning brief or an end-of-day report. You can review the existing jobs or create a new one with a name, a cadence, and detailed instructions. You can also just ask the employee on Slack to run something daily, and the job shows up here.

Workspace files

Workspace files are the configuration files generated for the employee, such as their agent profile, user details, and heartbeat. You can read and edit them here, and they stay in sync with the employee's local instance. Editing them directly is best left to technical users, since a small change can alter behavior in ways that are hard to trace.

Note

If cln.work manages your AI employee's infrastructure, reach out before hand-editing workspace files and we can make the change with you.

Documents

As employees work, they produce documents: research, drafts, analyses, emails. They all collect here, where you can read them and edit any section yourself. After you make an edit, let the employee know so they pick up your changes.

Was this page helpful?