How Grok Bot routines run

Short answer: A routine does two things. It assigns a workflow to one Bot, and it tells that Bot when to run. The when can be a schedule or, where supported, after an event. A skill describes how to do the task. A routine decides who runs it and when. The docs say to test the skill on a real one-time task before turning it into a routine. Closing the app or the computer does not stop it. After a long time away, if you do not reply, the product may ask whether to keep routines running and pause them when there is no response. (official FAQ, skills and routines)

How a routine relates to a skill

The FAQ distinction is short. A skill describes how to perform a task. A routine assigns a workflow to one Bot and tells it when to run, on a schedule or, where supported, after an event. Test the skill on a real one-time task before you automate it. (official FAQ)

The skills-and-routines page writes the order in full. Start with a one-time task. Make it reliable, save the method as a skill, and only then automate it. A skill is a reusable set of instructions. It captures steps, decision rules, the expected output and safety boundaries. Skills are available across your Bots, although a Bot may still need the relevant connector or login to use one. (skills and routines)

A useful skill states six things: when to use it, the required inputs and access, the sequence of work, how to validate the result, what to return, and what requires approval. You can ask the owning Bot to save the process you just used, and to include the source systems, the output format and the approval rule. (skills and routines)

Private skills are one library shared by all of your Bots. On desktop, type / in the composer to reference a saved skill. Type @ for Bots, groups, routines and connectors. If a skill does not appear in the slash menu, open Connect apps, Your plugins, Manage plugins and skills, and check Private skills. (skills and routines)

When Teach a task is available, you can record a browser workflow from the computer view instead of typing every step. The documented steps are: open a one-to-one Bot conversation and its computer view, choose Teach a task, describe the result you are about to demonstrate, perform the workflow once, stop the recording, review the skill the Bot creates, and test it on a safe example before you schedule anything. The recording covers visible computer interaction for up to ten minutes. It does not record microphone audio. Use the secure handoff for credentials, and do not expose secrets in the demonstration. The learned skill is a draft. Add decision rules, failure handling and approval boundaries that one example may not show. The control may roll out gradually. If you do not see it, ask the Bot to create a skill from written instructions and the completed task. (skills and routines, official FAQ)

The one-line definition is the routine entry in the glossary. This page is about how a routine runs.

On a schedule, or after an event

1. Make one real task reliable. The use-cases page starts with read-and-prepare work, then a review, and only then a routine. Do not put a schedule on a process that is not stable yet. (use cases, get started)
2. Save the method as a skill. State when to use it, the inputs, the sequence, how to validate, what to return, and where approval is required. Test it on a second input. (skills and routines, use cases)
3. Tell the owning Bot when to run. Confirm the owner, the schedule and time zone, the input source, the expected result, the approval boundary, and what should happen when a source is missing. The Bot creates the routine and shows the next run. (skills and routines)
4. Check it with a test run. Use Test run after you create or edit a routine. A test run performs real work. It can navigate websites, change files and call connected tools, so use safe inputs and keep writes behind approval. (skills and routines)
5. Define failure, then leave it on the schedule. The use-cases page says to create a routine only when retries and failure cases are defined. Have a policy for no data and for stale data. (use cases, skills and routines)

The schedule example in the docs is a message to the Bot that owns the job: every weekday at 8:00 AM, run a named skill against the current list, post a linked watch list in this conversation, do not contact customers, and if the source data is unavailable, report the failure instead of using old data. What you confirm is those six points, not a vague “run this daily.” (skills and routines)

An event trigger is a second path, and it is not available everywhere. The docs say Cursor account integrations can start a routine from an event, such as a Slack message or a GitHub notification. Those integrations are separate from Slack or GitHub plugins and may need their own connection flow. Keep the matching rule narrow and the response clear. The documented example fires only when a message in a named channel contains both a support-ticket link and a set phrase. The Bot then opens the ticket, reproduces the issue in staging, and posts a repro pack in the conversation. It does not post back to Slack without approval. A broad listener such as “every new message” creates noise, consumes usage, and raises the chance of acting on irrelevant input. (skills and routines)

Schedules have a hard gap. Routine schedules must be at least five minutes apart. A Bot can own up to 50 routines. The app keeps the 20 most recent run records for each routine. (skills and routines)

Background routines can run while the laptop is closed. That is not the same as someone watching the window. (skills and routines)

Does closing the app stop it?

Closing the app does not stop it. The FAQ says work runs on the cloud computer, and closing the app, laptop or phone does not stop a background turn or a routine. (official FAQ)

Hiding the Bot from the sidebar does not pause it either. The Bots docs say Hide from sidebar only removes the Bot from the main list. It does not delete the work. Hiding does not pause the Bot or its routines. (Bots docs)

It stops when you pause or delete it. Open the Bot, choose View conversation details, then open Routines. You can see its routines and recent runs, enable or pause one, run a test, edit the schedule or the instructions, inspect recent success and failure, and delete it. Deleting a routine is immediate and has no undo. Deleting a Bot also removes routines owned by that Bot. (skills and routines, Bots docs)

There is another pause that is not caused by closing the app. To control unattended usage, Grok Bot may ask whether to keep routines running after a long period away, and pause them if there is no response. Review paused routines when you return. (skills and routines)

On a phone you can look, but editing and testing currently happen on desktop. The mobile docs say to open a Bot’s profile to review its routines: schedule, next run, instruction and Run history. Active pauses or resumes a routine, and you can delete one from the profile. Editing the schedule or the instruction, and testing a routine, currently require the desktop app. (mobile docs)

When you add someone else’s template, a public link shows the routines. That is part of the recipe, not the copy that is running on the author’s account. The steps are in how to add a template. How to notice whether a template runs on its own is in how to choose a template. The template guide also describes routines as the triggers that make a Bot run on its own, such as how often it runs. (Bots docs, x.ai template guide)

What not to hand to a routine

The docs put the automatable part in preparation, not in unsupervised execution. Have the Bot draft, reconcile or recommend first. Require approval for sending, purchasing, deleting, publishing or changing production systems. (skills and routines)

The use-cases page turns that into rules for roles. Sales outbound can become a nightly research routine after the output is reliable, and it still stops at the review list. It does not send or enroll anyone. Paid-media changes stay behind approval even after the analysis is a routine. Product performance can use a routine for a recurring health report, not for unsupervised production changes. A chief-of-staff digest is tuned by marking what was useful and what was noise, then scheduled for a time you can review it. (use cases)

A missing source is not something memory should fill in. The docs say to include a no-data and stale-data policy, to make retries idempotent where possible, and to tell the Bot where to report partial completion. Re-test after a website, connector or source format changes. (skills and routines)

Before you trust a test run, review five checks. Did it select current inputs? Does the output meet the required format? Does every action have a source or an audit trail? Did it stop at the intended approval point? Are failure states explicit? (skills and routines)

The security docs list changes to routines and event triggers among the automation writes Auto Review can look at. Changing a routine can itself be an action that needs review, not only the actions the routine takes later. Personal Ask first rules live under Settings, General, Auto-review, and they should stay narrow. Do not use a broad rule such as allowing everything in the browser for sending, publishing, deleting, purchasing or production changes. How to set that is on the template safety checklist. (security docs, official FAQ)

When a project or login should no longer be available, the cleanup list starts by pausing or deleting related routines. Then sign out of websites on the computer, uninstall plugins and revoke them in the source service, and remove sensitive files from /workspace. Deleting a Bot does not remove files or browser sessions on the computer. (security docs)

This site groups templates whose summaries watch for change in the Monitoring & alerts collection. That is a directory cut, not an official routine catalog. Whether a template runs on its own is still whatever its preview lists under routines.

FAQ

Does closing the app or the laptop stop a routine?

Not by itself. The FAQ says closing the app, laptop or phone does not stop a routine, because the work runs on the cloud computer. It does stop if you pause or delete it, if you delete the Bot, or if you have been away for a long period, the product asks, and you do not reply. Hiding a Bot does not pause its routines.

Are a skill and a routine the same thing?

No. A skill describes how to perform a task, and your Bots can share that library. A routine assigns a workflow to one Bot and says when it runs, on a schedule or, where supported, after an event. The docs say to test the skill on a real task before you automate it.

What belongs on a schedule, and what does not?

Drafting, reconciling and recommending are what the docs automate first, and a missing source should be reported as a failure rather than filled with old data. Sending, purchasing, deleting, publishing and changing production systems should stop for approval. A broad “every new message” trigger also creates noise and irrelevant action, and the docs say to avoid it.

Based on the official FAQ (Sep 29, 2026), skills and routines, Bots and get started (Oct 12, 2026), use cases, mobile, security and the x.ai template guide (Sep 8, 2026). Last checked: 2026-10-18. Bests.one is not affiliated with xAI.