by /
We published a week of Grok Bot pages that named the roster and skipped the two buttons that actually make a Bot an employee: skill and routine. xAI already used our words. We had not put them on the map.
Stale/build-agents/ still said 7-day /loop
Gluedteach-a-task ≠ routine
SameClaude scheduled task = Grok routine
What we were asked
Dennis sent the xAI skills and routines page and said: account for this in our Grok documentation, given what we already published about skills and scheduled tasks via Claude. Harmonize them. Update Grok / Cursor / Grok Bot so we practice what we preach. We had shipped a pile of Grok Bot articles in the last two days.
What we found
xAI’s page is not a new religion. A skill is a reusable how. A routine tells one Bot when to run it — on a schedule or, where supported, after an event. Start with a one-time task. Make it reliable. Save the method. Only then automate. Test run is real work. 50 routines per Bot. 20 run records. Laptop closed is allowed. Teach-a-task writes a skill draft, not a clock.
We had already published that ladder:
- A skill is a function. An agent is a person.
- Persistent agents — the job is a schedule, a QA cycle, and working files.
- 18 scheduled tasks and how to use Claude — Cowork scheduled tasks, CCS, save a Skill then put it on a clock.
- The token-usage dashboard — a worked Cowork scheduled task with a Run-now to pre-approve tools.
What was missing was the vendor dictionary. /grok-bot/ explained the roster versus Heavy’s four-at-once reply, then one table row glued “show it once” to “save a routine.” The hub said routines live on the computer and never taught Settings → Plugins, slash-menu enablement, Test run, or View conversation details → Routines. /build-agents/ still told the world that consumer Grok has Automations and Grok Build /loop expires in seven days, so long-lived production scheduling needs another host. That sentence is true of grok.com Build. It is false of Grok Bot.
What we changed
| Surface | Change |
|---|---|
| New map | /skills-and-routines/ — Claude scheduled task = Grok Bot routine = Cursor Automation. Paste blocks in all three clothes. |
| Desk SOP | /grok-bot/ — unglued teach vs routine; linked the xAI page; mechanic checklist now includes save-skill-then-schedule. |
| Hub | how I use Grok Bot — first-hour and setup steps for Plugins, slash menu, Test run, Routines pane. |
| Job rung | persistent agents FAQ: what vendors call the schedule. |
| Runtime map | /build-agents/ scheduling row: Grok Bot routines, not 7-day /loop as the production host. |
| Cursor | Personal skill skill-then-routine plus a user rule. Same order: how, then when. |
| Packs | DealCon-Skills/skill-then-routine.md mandated for the next daily pack run. |
We did not enable a stack of new Grok Bot routines on 18 Aug. Live spending on this Ultra seat, early afternoon: Cursor Models 11%, Other Models 0%, Grok Bot weekly 100% until 19 Aug, on-demand $358.49 / $380. The 7:00am Cursor email said Other Models were exhausted and I should switch to Grok 4.6. Default already is Grok 4.6. The 100% bar is Grok Bot. Practicing what we preach is: write the how, paste the when, wait for a buffet that is not already grocery.
What we got wrong earlier this week
On 17–18 Aug we split Grok Bot from Heavy’s four-at-once reply and from grok.com chat. That split was right. We then staffed the public roster as if naming desks was the job. A named Bot with no skill and no routine is still a chat window with a photo. xAI’s own overview said you can turn a finished path into a reusable skill or a proactive routine. We linked the overview and did not follow the next page.
Cost
This pass ran on Cursor Grok 4.6 (workshop / Cursor Models), not on the Grok Bot weekly meter. No Extra Credits spent to “test” twelve new routines.
Where this sits
Receipt for the glossary. The SOP is /skills-and-routines/.
Grok Bot desk · Persistent agents · How we mapped the usage buckets
Meta article, 18 Aug 2026. Built per the meta article prompt against /skills-and-routines/.
Dennis Yu
Dennis Yu is the CEO of Local Service Spotlight, a platform that amplifies the reputations of contractors and local service businesses using the Content Factory process. He is a former search engine engineer who has spent a billion dollars on Google and Facebook ads for Nike, Quiznos, Ashley Furniture, Red Bull, State Farm, and other brands.
Dennis has achieved 25% of his goal of creating a million digital marketing jobs by partnering with universities, professional organizations, and agencies. Through Local Service Spotlight, he teaches the Dollar a Day strategy and Content Factory training to help local service businesses enhance their existing local reputation and make the phone ring.
Dennis coaches young adult agency owners serving plumbers, AC technicians, landscapers, roofers, electricians, and believes there should be a standard in measuring local marketing efforts, much like doctors and plumbers must be certified. He has appeared on 353 podcasts with 619 credited episodes — see the full list of his podcast appearances.