Grok Bot routines and skills

Skills capture how to do a task. Routines tell a Bot when to run that workflow — on a schedule or, where supported, after an event. Automating before the process is reliable multiplies token and computer-use cost.

Hub: Tokenminning in Grok Bot.

Build in this order

Complete a one-time task successfully
Save the method as a skill with inputs, steps, validation, and approval boundaries
Run the skill manually on a second input
Create a routine only when failure cases and stale-data handling are defined
Test run with safe inputs before enabling unattended execution

Save a useful skill

Ask:

Save the process we just used as a skill called “Weekly account health.” Include source systems, risk definitions, output format, and the rule that customer contact always requires approval.

A useful skill states:

When to use it
Required inputs and access
The sequence of work
How to validate the result
What to return
What requires approval

Enable skills per Bot under Settings → Plugins → Yours. Type / in the composer to reference a saved skill.

Schedule routines narrowly

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: owning Bot, schedule and time zone, input source, expected result, approval boundary, and missing-source behavior.

Event triggers — keep them tight

Cursor account integrations can start a routine from Slack or GitHub events. Define a narrow match:

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

Avoid broad listeners such as “every new message.” They create noise, consume usage, and increase irrelevant computer-use.

Limits and pause controls

Limit	Value
Routines per Bot	Up to 50
Recent run records	20 per routine
Teach-by-demonstration	Up to 10 minutes of visible interaction

Grok Bot may ask whether to keep routines running after a long absence and pause them without a response. Review paused routines when you return.

Manage routines: open the Bot → View conversation details → Routines — enable, pause, test, edit, or delete.

Design for trust (and lower burn)

Automate preparation before execution — draft, reconcile, recommend first
Require approval for sending, purchasing, deleting, publishing, or production changes
Include a no-data and stale-data rule — do not guess from old exports
Make retries idempotent where possible
Re-test after a website, connector, or source format changes

Teach by demonstration

When Teach a task is available, demonstrate a browser workflow once instead of writing every step. The learned skill is a draft — add decision rules and failure handling the demo may not show. Avoid exposing secrets during recording; use secure handoff for credentials.