Grok Bots Directory
Submit a Grok Bot

Chief of staff: the one Grok Bot you talk to, layers that delegate

A chief of staff Grok Bot is the one bot you talk to in a layered bot team. For a big project you plan with it, and each layer below splits the work further until Cursor cloud agents do it.

lauren @poteto

Source: Peter Yang podcast ·

The playbook

The playbook is a chain of Grok Bots in which only the bottom layer does the work. lauren mostly talks to her chief of staff, and the chief of staff talks to her eng lead, Matcha. She had Dr. Eggbot set up the eng lead so that its job is to break tasks into smaller pieces, delegate them, and supervise other bots. It does not do the work on its own. Its description names the engineer bots by bot ID and tells it to always delegate to them.

The engineer bots also stay out of the work. For each task they spawn a Cursor cloud agent, which you can open in Cursor or on the web. In her pstack guide, lauren says this frees the bot for other things and keeps its context window clean: "I think of my bots more as coordinators who manage and supervise cloud agents." Cloud agents also let her pick the model and reasoning level for each task: a fast model for simple tasks, more reasoning for tasks that need planning.

How a Grok Bot team runs the layers

  1. Hand the project to the chief of staff

    You talk to one bot. For a large project, plan first in a Notion doc. The chief of staff then works out the phases and hands them down to the eng lead.

  2. The eng lead splits and delegates

    It breaks each phase into smaller pieces, gives them to named engineer bots, and supervises. Its description tells it never to do the work itself.

  3. Engineer bots spawn cloud agents

    Each engineer bot starts a Cursor cloud agent for its piece and picks the model and reasoning level for that task. The bot supervises and does not code.

  4. Cloud agents do the work and show proof

    lauren's prompt in her pstack guide has the agent build with /poteto-mode, check the change with the project's verification skill, and return a video and screenshots as proof.

  5. Pass context bot to bot

    Because every layer is a bot, you can point one bot at another. lauren can tell Matcha to go talk to her social bot about a bug reported on X and come back with a design proposal, without pasting the context in again.

Inspired templates11

A template is listed here only when coordinating other bots or agents is its own job: its own skills or memories split the work and hand it to specialists or cloud agents instead of doing it. Chiefs of staff that handle inbox and calendar work themselves are not listed.

All templates with pstack

Orchestrator

Orchestrator

by index

Runs GitHub engineering through mattpocock dual-track docs/code PRs: Spec and tickets on a docs draft, then Cloud Agent implementers on a clean code draft — wi…

  • Memories8
  • Skills1
  • Routines1
Lingxi's Engineer Bot

Lingxi's Engineer Bot

by Lingxi

A hands-off engineering supervisor. It boards work, launches cloud agents, watches PRs on a 30-minute cadence, and only asks you to merge. For anyone who wants…

  • Memories24
  • Skills1
  • Routines0
Jarvis

Jarvis

by Andrew

Orchestrates a roster of specialist Grok Bots. Routes work, packages attention briefs, and only pulls you in for judgment, money, publishing, or a blocked bot.

  • Memories3
  • Skills3
  • Routines2
The Boss by David

The Boss by David

by David

Takes your goal, breaks it into tasks, delegates each to the right specialist, and assembles the final result. It doesn’t do the deep work itself — it orchestr…

  • Memories3
  • Skills0
  • Routines0
Engineering Lead

Engineering Lead

An engineering lead that owns the loop: breaks work into Cloud Agents, chases CI, reviews, and stalls on a weekday cadence, and only pings you when something i…

  • Memories3
  • Skills5
  • Routines0

More playbooks

All playbooks