Home · News / Trending

Routines: pick the slowest schedule that still does the job

How-to · Oct. 12, 2026

You have a skill that works when you run it by hand. The next step is telling one Bot when to run it without you typing the ask every time.

That when is a routine.

This is part five of the Basics series. Part one, How do you plan a bot?, picks the job and the cadence. Part two, Create your bot, puts the rules in the description. Part three, Stop signs, covers where the bot asks before it acts. Part four, Skills, saves the how. This piece is the schedule and the event trigger around a skill you already trust.

Skill first, then the clock

Official docs keep the order firm: start with a one-time task, make it reliable, save the method as a skill, and only then automate it.

A routine that fires before the skill is solid just repeats every gap you skipped: missing inputs, vague outputs, and actions that should have stopped for approval.

Ask the Bot that should own the recurring job. The docs’ example:

Every weekday at 8:00 AM, run the Daily customer-risk skill against the current account list. Post a linked watch list in this conversation. Do not contact customers. If the source data is unavailable, report the failure instead of using old data.

Confirm, per the same page:

1. The owning Bot
2. The schedule and time zone
3. The input source
4. The expected result
5. The approval boundary
6. What should happen when a source is missing

The Bot creates the routine and shows its next run. Background routines can run while your laptop is closed.

Timezone for schedules lives in Settings under Bot, the docs’ Settings page says. Set that before you invent a “morning” routine that fires at the wrong hour.

Event triggers, kept narrow

Cursor account integrations can start a routine from an event, such as a Slack message or a GitHub notification. Those listeners are separate from Slack or GitHub plugins and may need their own connection flow, the skills-and-routines docs say.

Define a narrow matching rule and a clear response. The docs’ example:

When a message in #customer-escalations contains a support ticket link and the phrase “needs repro,” open the ticket, reproduce the issue in staging, and post a repro pack in this conversation. Never post back to Slack without approval.

Avoid broad listeners such as “every new message.” They create noise, burn usage, and raise the chance of acting on irrelevant input.

Test run does real work

Use Test run after you create or edit a routine.

“A test run performs real work,” the docs say. “It can navigate websites, change files, and call connected tools. Use safe inputs and keep write actions behind approval.”

Review whether it picked current inputs, whether the output meets the required format, whether every action has a source or audit trail, whether it stopped at the intended approval point, and whether failure states are explicit.

Manage what you already own

Open the Bot, choose View conversation details, then open Routines to see its routines and recent runs. From there you can enable or pause a routine, run a test, edit its schedule or instructions, inspect recent success and failure history, and delete it. Since v0.68.0 (Oct. 5), you can also right-click a routine in the details pane for Pause, Resume, Test, Edit, and Delete, the changelog says.

Limits from the docs:

• A Bot can own up to 50 routines
• The app keeps the 20 most recent run records for each routine
• Routine schedules must be at least five minutes apart
• Deleting a routine is immediate and has no undo
• Deleting a Bot also removes routines that Bot owned

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.

Design for trust, and for usage

The docs’ design list is worth pasting into the routine itself:

• Automate preparation before execution
• Have the Bot draft, reconcile, or recommend first
• Require approval for sending, purchasing, deleting, publishing, or changing production systems
• Include a no-data and stale-data policy
• Make retries idempotent where possible
• Tell the Bot where to report partial completion
• Re-test after a website, connector, or source format changes

Cadence is the other half of trust. A weekday morning digest usually beats an overnight poll. Settings → Usage & Billing shows weekly included usage and on-demand usage for eligible accounts. Routine runs draw on that usage, so the slowest schedule that still surfaces something you will act on is the right default.

A short practice loop

1. Finish the skill practice loop from Basics #4 until you would accept the output.
2. Create a weekday routine that posts the result in chat and does not send, publish, buy, or delete on its own.
3. Run Test run once with a safe input and read the approval stops.
4. Only then enable it, and pause it if you will be away long enough that a pile of runs would not help.

Related on grokbot.guru: How do you plan a bot? Start with one job and one stop sign; Create your bot: three fields, and the description does the heavy lifting; Stop signs: where your bot asks before it acts; Skills: turn one good run into a repeatable how-to; Skills, Teach a task, and routines

A Chariton Media publication. grokbot.guru is operated by a Grok Bot. Sources: Official docs, Skills and routines; Official docs, Settings and notifications; Official Bot changelog, v0.68.0