Run all of this inside the full Agent OS — join the AI Profit Boardroom →

agentos.guide › blog

Grok Bot vs Copilot: How To Run GitHub Copilot and Grok Bot as One Crew In Your Agent OS, With 5 Copy-Paste Prompts

By Julian Goldie · 7 November 2026 · agentos.guide

Illustration of a code editor window and a cloud computer with app icons, linked by a gold line through a central hub on a dark aubergine background

Grok Bot vs Copilot stops being a contest once both sit inside your Agent OS: GitHub Copilot owns the code, Grok Bot owns the errands around it, and your own agent writes the briefs for both.

Copilot here is GitHub Copilot, the coding product, not the Microsoft 365 one.

Copilot works in your editor, your terminal and your GitHub repositories.

Grok Bot works from one cloud computer that all your Bots share, each with its own screen.

This guide gives you a quick start, five prompts to paste, and the daily rhythm I would run.

Want my team to plan the split for your business? Get your Backlink Gameplan, or try Polar Rank free for 3 days.

Who gets which job

JobGive it toWhy
Change, test and review codeCopilotLives in the repo and returns pull requests
Turn a client request into a clear issueYour Agent OS agentIt knows your clients and your house style
Inbox, calendar and website choresGrok BotConnected apps plus a browser on its cloud computer
Plain-English reports on repo activityGrok BotIts GitHub connector reads commits and comments
Merging, paying, sendingYouKeep the final click on anything you cannot undo

This split is my own reading of each vendor's docs, as neither GitHub nor xAI compares the two.

Quick start in five moves

  1. Install Copilot in your editor, or install Copilot CLI on Linux, macOS or Windows, per the CLI docs.
  2. Turn on the Copilot cloud agent (formerly the Copilot coding agent) for one private repo.
  3. Install the Grok Bot desktop app and sign in with your Cursor account, as the Grok Bot FAQ explains.
  4. Create one Bot with one job and connect only the apps that job needs.
  5. Add Ask first rules before you give either side anything that leaves the building.

Grok Bot comes with Premium, Super and Ultra on The X Subscription, per the plans page; Plus does not include it.

Every Copilot plan includes Copilot CLI and the Copilot app, according to GitHub's plans page.

Prompt 1: write the Copilot house rules

Copilot reads a repository file at .github/copilot-instructions.md and also understands AGENTS.md, per GitHub's custom instructions page.

Ask your Agent OS agent to draft it from what it already knows about the project.

Read this repo's README, package files and test setup.
Draft .github/copilot-instructions.md for GitHub Copilot, under 60 lines:
1. How to install, build and run the tests.
2. Folders Copilot must never edit.
3. Our code style in five bullet points.
4. What every pull request must include: a summary, test results, screenshots for UI changes.
Do not commit it. Show me the file first.

Review it, commit it, and every Copilot surface on that repo picks up the same rules.

Prompt 2: turn a client request into an issue Copilot can take

When you assign an issue to Copilot, the issue assignment docs say it reads the title, the description and existing comments, then stops reading new comments.

So the issue has to be complete before you assign it.

Here is a client email: [paste it].
Write a GitHub issue for Copilot cloud agent:
- Title under 70 characters.
- What to change, which page or file, and what "done" looks like.
- Acceptance checks I can verify in the preview.
- Anything Copilot must not touch.
If the request is vague, list the questions I should ask the client instead.

Paste the issue into GitHub, assign it to Copilot, and steer any later changes on the pull request rather than the issue.

🔥 Want the whole two-agent Agent OS set up for you?

Inside the AI Profit Boardroom you get step-by-step video trainings, the installable Agent OS, four coaching calls a week and 3,400+ members running crews like this one.

→ Join the AI Profit Boardroom

Prompt 3: brief a Grok Bot with a job and hard limits

xAI's approvals page tells you to state in the request which actions a Bot may take and where it must stop.

Paste this into Grok Bot itself, not your Agent OS.

Your job: client communications for our web agency.
Every weekday at 8am, read new client emails in Gmail.
For each one, draft a reply in our tone and label it Bug, Change request or Question.
For Bug and Change request, write a short summary I can turn into a GitHub issue.
Never send an email, never post in Slack, never buy anything.
Leave every draft for me and post a short list of what you drafted.

Then open Settings, General, and Custom Rules under Bot, and add Ask first rules for "send an email" and "post in Slack" so the limits hold even if the brief is ignored.

Prompt 4: a routine that watches GitHub and reports in Slack

The skills and routines page lets a routine start on a schedule or on a GitHub, Slack or email event, among others.

The Grok Bot changelog says the GitHub connector can list repos, compare commits and read pull request comments (V0.75.2).

Every Friday at 4pm, use the GitHub connector to read pull requests merged this week in [repo].
For each one, write one plain-English line a client would understand.
Group the lines by client.
Draft the summary as a Slack message to #client-updates and wait for my approval.
Save this as a skill called "Friday repo summary" before you schedule it.

Run it once by hand, fix the output, save the skill, and only then let the routine run.

Prompt 5: the hand-back from Slack to Copilot

GitHub says @GitHub in Slack and Teams became generally available on 6 November 2026 and can create issues and open pull requests from a conversation.

That gives your team a relay: the Bot's summary lands in Slack, and a developer replies in the thread when something needs code.

@GitHub create an issue in [repo] from this thread.
Title: [short title].
Body: the client's request above, the acceptance checks from my last message, and "do not edit /config".
Then implement it and open a pull request for my review.

Check the repository it picked before you send, because GitHub now guesses the repo from your message.

The daily workflow

  1. 8am: the Bot drafts client replies and summaries; you approve the drafts.
  2. Mid-morning: your Agent OS agent turns the change requests into full issues with Prompt 2.
  3. Midday: you assign the issues to Copilot, which works in a GitHub-hosted sandbox.
  4. Afternoon: you review the pull requests and merge the good ones.
  5. Friday: the routine posts the week's plain-English summary for approval.

Safety checks before either one runs alone

📞 Want a second pair of eyes on your crew?

Get your Backlink Gameplan and my team will check your lanes, your briefs and your Ask first rules, or try Polar Rank free for 3 days.

Backlink Gameplan or try Polar Rank free for 3 days

Example: a three-person web studio

The studio keeps every client site in its own GitHub repo with a house rules file from Prompt 1.

Each morning a Bot sorts client email and drafts replies, while the owner approves them over coffee.

Change requests go through Prompt 2 and land on Copilot, and the developer reviews each pull request before lunch.

On Friday the routine from Prompt 4 drafts the client update, and the owner approves it in Slack.

For the same idea with a different coding tool, see Grok Bot vs Cursor in your Agent OS.

Mistakes that cost you a week

Giving both agents the same repo job is the first one, because you end up reviewing two sets of changes to one file.

Assigning a half-written issue is the second, since Copilot will not see the comment you add ten minutes later.

Letting a Bot store a client login as a plain file on its computer is the third, as every other Bot on your account can read it; use the Vault or a Secure Form instead.

Picking Actions for a session when you wanted connector tools is the fourth: the cloud sandbox docs say you choose Cloud or Actions at the start and cannot switch.

Skipping the dry run is the last, so always run a new routine by hand once before you schedule it.

Which one to start with

Start with Copilot if code is your bottleneck and your repos already live on GitHub.

Start with Grok Bot if your inbox and admin are the bottleneck and nobody on the team writes code.

Add the second lane only when the first one has run cleanly for a fortnight.

Paste Prompt 1 or Prompt 3 today, whichever matches your bottleneck.

🔥 Get the prompt library behind this guide

Join the AI Profit Boardroom for step-by-step video trainings, the installable Agent OS, four coaching calls a week and 3,400+ members.

→ Get inside the Boardroom

FAQ

Can my Agent OS agent assign issues to Copilot directly?

This guide keeps that step with you: the agent drafts the issue, and you paste and assign it on GitHub or ask @GitHub in Slack to create the issue and open a pull request.

Where do I paste Prompt 3?

Into Grok Bot itself, as the brief for one Bot. Then add Ask first rules under Settings, General, Custom Rules so the limits hold.

Does Copilot read comments I add after assigning an issue?

No. GitHub says Copilot stops reading new issue comments once it is assigned, so put later changes on its pull request.

Do my Bots each need their own computer for this workflow?

No. All your Bots share one cloud computer and each one gets its own screen, so the email Bot and the reporting Bot can work at the same time.

Which should I set up first?

Whichever matches your bottleneck: Copilot for code on GitHub, Grok Bot for inbox and admin work.

Final word on Grok Bot vs Copilot

Pick your first lane tonight, paste the matching prompt tomorrow, and get your Backlink Gameplan if you want help running grok bot vs copilot as one crew, or try Polar Rank free for 3 days.

About Julian

I'm Julian Goldie, an AI entrepreneur, SEO expert and the founder of the AI Profit Boardroom, which has 3,400+ members.

I help business owners scale with AI agents, automation and SEO.

My YouTube channel has 400,000+ subscribers, and I run Goldie Agency, a seven-figure SEO agency.

→ Get my best AI training inside the AI Profit Boardroom

Related guides on agentos.guide

→ Grok Bot vs Codex: How To Run Both As One Crew In Your Agent OS, With 5 Copy-Paste Prompts

→ Grok Bot AI Full Course (Step By Step)

→ The Ghost Team

→ Hermes Agent Browser: What It Is, The Quick-Start And Where It Fits In An Agent OS

Get the whole Agent OS

The installable system behind every guide on this site — dashboard, agents, memory and pipelines — updated as new tools land, with 3,400+ members and four coaching calls a week.

Join the AI Profit Boardroom →

📺 Video notes + links to the tools 👉

🎥 Learn how I make these videos 👉

🆓 Get a FREE AI Course + Community + 1,000 AI Agents 👉

← all guides on agentos.guide