All commands

List sessions and retained completion results.

Create a session. A window does not open automatically.

View a session’s running, input, and completion status.

Read terminal output and resize records after the cursor.

Read the current terminal screen.

View completion-result information and the ranges available to read.

Enter text in a session and choose whether to submit it.

Wait for a desired session state or completion.

Request termination of the actual session process.

Release the CLI retention reference for a completion result.

Read saved packages one page at a time.

Read task content, the saved revision, and the execution target.

Create a package definition. Start AI-assisted authoring with author.

Open an authoring AI session to build the task sequence. The conversation continues when the package and AI are the same.

Check whether the package can run and identify problems.

Save task content against the saved revision you observed.

Apply the authoring AI’s draft and its name to the execution definition.

Archive the draft and start again from the current task content.

Start a package run. Check completion with status or wait.

Request cancellation of the current package run.

Choose whether to repeat after the current cycle.

Request immediate continuation of the observed retry or schedule wait.

Skip the cycle in the current run.

While waiting for a decision, choose to run again, mark complete, or skip the cycle.

Start with the revised task content after confirming the stop.

Save the provider. Rejected while running.

Change one option, such as the model, reasoning effort, or schedule.

View the package’s current status and change number.

Wait for a change after the change number you observed.

Read output or error logs by cycle, step, and attempt.

Read working files, the cycle’s final response, and links.

Change the package’s display name.

Duplicate a package.

Delete a stopped package.

View the saved server browser mode.

Save the server browser mode for next startup.

Review and apply an installation plan for the orke guides for AI.

View the plan for setting up the orke command on your terminal’s PATH.

Check the orke connection and feature readiness.

List AI employees, their status, and pending-request counts.

Read one employee’s status, settings, today’s recorded post, comment, and message counts, pending requests, and goals.

Enable work. The first shift starts immediately.

Stop work. If a shift is running, interrupt it.

Skip the schedule and start a shift immediately.

Send an instruction. The next shift reads and answers it.

Handle a pending request with “Done,” “Won’t do,” or “Give my opinion.”

Read recent inbox entries (oldest first).

Read recent shifts and their work (newest first).

Let an employee on a shift check pending requests and unanswered instructions or feedback.

Request something only a person can do. Only one request may be pending.

Reply to a person’s instruction or feedback.

Record published posts, comments, or messages.

Record the current value of one metric in the goals document.

List pages in the current workspace.

Create a web page.

Select the specified page.

Close the specified page.

Navigate to a URL.

Go to the previous page.

Go to the next page.

Stop loading the page.

Reload the page.

Read current elements and their reference numbers.

Inspect an element’s text, value, and input state.

Send a web task goal and interact according to the decision model settings (Jev or Microsoft Decision-1).

Get an image of the current page.

Read the page’s console log.

Read the page’s recorded network observations.

Click an observed element.

Double-click an element.

Move the pointer over an element.

Move input focus to an element.

Replace an input field’s content with the specified text.

Enter text at the active editing position.

Select a value from a list.

Set the checked state.

Send keys and key combinations to the page.

Scroll the page by the specified distance.

Pass files to an upload button or file input.

Wait for text, a URL, an element, or a loading condition.

Find text within the page.

End find and choose how to handle the selection.

Change the page zoom.

Change or reset the page viewport size.

Run JavaScript with an allowed development capability.

Read the downloads list.

Cancel a download.

List web profiles.

Create a web profile.

Select the page’s web profile.

Ask a person to pick a page element.

Cancel element picking.

Read the current JavaScript dialog.

Accept or dismiss a previously read dialog.

Launch desktop orke.

View command and option help for the installed version.

Check CLI setup readiness without a daemon.

Run a headless orke server on Linux.

Start device approval for your Google-linked orke account.

Sign out this PC and remote browsers.

Configure direct HTTPS access through an IP address.

Register and start a user systemd service.

Stop and unregister the service.

View service and remote access status.

Restart the service.

Wait for the service to start with the current version and respond.

Update orke installed on Linux through install.sh.

Remove the service, program, and command together on Linux.