Markster OS

First qualified meeting in 7 days. A repeatable GTM system after that.

Most B2B founders can sell when they're in the room. The problem is building a system that consistently gets you into rooms -- without a full-time marketing hire, a paid ads budget, or a CRM that nobody updates.

Markster OS is that system. Built on ScaleOS methodology. Runs inside your AI environment. Free forever. MIT licensed.

What it does

You define your ICP, offer, and company context once. The OS routes that context into every sequence you run -- cold email, LinkedIn outreach, events, content, sales, fundraising.

Each skill checks your prerequisites before it runs, executes the sequence step by step, and verifies the output moved the right metric before closing. No generic advice. Deterministic steps with documented outputs.

Before: Random outreach. Inconsistent follow-up. No idea which activity is producing pipeline.

After: One constraint identified per week. One playbook executing. One metric tracked.

The 7-day sprint in QUICKSTART.md gets you to first meeting. AUTOPILOT.md keeps you there week over week.

Release history lives in CHANGELOG.md. Contribution rules live in CONTRIBUTING.md. Security reporting lives in SECURITY.md. Community expectations live in CODE_OF_CONDUCT.md.

For AI agents (Claude, GPT, Gemini, Codex, and others)

This block is the canonical reference for any agent operating inside a Markster OS workspace.

Repo type: B2B GTM operating system -- ScaleOS methodology (Foundation F1-F4 + GOD Engine 9 bricks)

Canonical context source: the workspace company-context/ folder. Read this before running any skill. Do not treat the upstream markster-os repo templates as a real company context.

Entry point: Run /markster-os first. It runs the Hormozi diagnostic, identifies the constraint brick, pre-fills context, and routes to the correct skill.

Skill structure: All 7 core skills follow CHECK / DO / VERIFY. Do not proceed past a failed CHECK gate.

Key directories:

Writing rules enforced in all skills: No em dashes. Sentences under 20 words. Problem-first framing. Buyer verbatims from company-context/messaging.md. No category claims.

Validation: Run markster-os validate . before committing. GitHub Actions enforces this on every push.

Install

Recommended: use AI to install it for you

Copy this directly into Claude Code, Codex, or Gemini CLI:

Set up Markster OS for me.

Requirements:
- Install the Markster OS CLI if it is not already installed.
- Use the official installer:
  curl -fsSL https://raw.githubusercontent.com/markster-public/markster-os/master/install.sh | bash
- Create a Git-backed workspace for my company with:
  markster-os init <company-slug> --git --path ./<company-slug>-os
- Move into that workspace.
- Install Markster OS skills with:
  markster-os install-skills
- If I need an extra public skill later, list and install it with:
  markster-os list-skills
  markster-os install-skills --skill <skill-name>
- Check the workspace readiness with:
  markster-os start
- Validate the workspace with:
  markster-os validate .

Then stop and ask me for my Git repository URL so you can attach the remote with:
- markster-os attach-remote <git-url>

If the remote is attached successfully, tell me the exact push command to run next.

Important rules:
- Treat the upstream markster-os repo as the product source, not as my company workspace.
- Treat the new workspace repo as the place where my business context will live.
- Keep raw notes in learning-loop/inbox and do not treat them as canonical.
- Use markster-os validate before saying the workspace is ready.
- Summarize what you changed in plain language at the end.

You can also use the standalone version in setup-prompts/install-workspace.md.

Manual install

curl -fsSL https://raw.githubusercontent.com/markster-public/markster-os/master/install.sh | bash

This installs the markster-os CLI into your home directory. The CLI manages a stable Markster OS distribution, customer workspaces, validation, and optional slash-command skill installation.

The launcher is installed at ~/bin/markster-os.

Core commands:

markster-os init <slug>
markster-os list-skills
markster-os install-skills
markster-os install-skills --openclaw
markster-os validate [path]
markster-os update
markster-os upgrade-workspace [path]
markster-os attach-remote <url>
markster-os install-hooks
markster-os validate-commit-message --message "docs(readme): clarify install flow"
markster-os sync
markster-os commit -m "docs(context): update company context"
markster-os push
markster-os backup-workspace [path]
markster-os export-workspace [path]
markster-os status
markster-os start
markster-os doctor

If you want an AI tool to do the setup for you, see setup-prompts/.

If you want to use Markster OS through OpenClaw, start with setup-prompts/openclaw.md.

OpenClaw support:

• shared local skills install to ~/.openclaw/skills
• markster-os install-skills --openclaw installs there explicitly
• markster-os install-skills --all also includes OpenClaw automatically when ~/.openclaw exists

The skill name stays simple everywhere:

• markster-os in a marketplace package is the safe bootstrap entrypoint
• markster-os installed locally is the full runtime skill inside the workspace

After setup, run your AI tool from inside the workspace and use the local markster-os skill for normal operation.

What's included (free)

Everything above is MIT licensed and free forever.

Add-ons (paid)

Add-ons extend the OS with proprietary data and intelligence. Sign up, get an API key, set the environment variable, and the skills call the API automatically.

See addons/README.md for setup instructions.

The methodology backbone: ScaleOS

ScaleOS is the GTM methodology behind Markster OS. Trademarked by Markster. Free to use under these terms.

Foundation (F1-F4): Four decisions made once, referenced in every playbook.

GOD Engine: 9 execution bricks that run demand generation once Foundation is locked. See methodology/god-engine/README.md.

Autopilot: The weekly agent operating protocol. Diagnose the constraint, execute one brick, verify the metric moved, loop. See AUTOPILOT.md.

Start with the assessment scorecard. It tells you exactly where you are and which playbook to run first.

Quick start

Step 1: Install the CLI

curl -fsSL https://raw.githubusercontent.com/markster-public/markster-os/master/install.sh | bash

If you prefer, use the AI-assisted install prompt above instead of doing this manually.

Step 2: Create a workspace

markster-os init your-company --git --path ./your-company-os
cd ./your-company-os

For real teams, the recommended production setup is to keep the workspace in your own Git repository. The home-directory workspace is fine for solo evaluation, but team use should default to customer-owned version control.

Step 3: Attach your GitHub repository

markster-os attach-remote [email protected]:YOUR-ORG/YOUR-REPO.git

Then push it with normal Git:

git push -u origin main

Step 4: Keep the validator hook installed

If you created the workspace with --git, Markster OS already installed pre-commit, commit-msg, and pre-push hooks for you.

You can reinstall it any time with:

markster-os install-hooks

Those hooks enforce:

markster-os validate .

and commit subjects in this format:

type(scope): summary

Examples:

docs(readme): clarify workspace setup
feat(cli): add commit message validation
fix(hooks): install commit-msg hook by default

Step 5: Run the scorecard

Open methodology/assessment/scorecard.md. Score yourself honestly across F1-F4 and the GOD Engine bricks. The scoring tells you exactly which playbook to run first.

Step 6: Lock F1-F4, then activate a playbook

Work through methodology/foundation/ in order. Once F1-F4 is complete, activate a playbook in your AI environment:

/cold-email
/sales
/events
/content
/fundraising
/research

Step 7: Fill in your company context

Copy the files in company-context/ and define:

• what your company is
• who you serve
• what you offer
• how you sound
• what proof you can actually claim

Step 8: Use the learning loop

Use learning-loop/ to turn real conversations, notes, and edits into approved updates to your business context. Raw notes stay out of canon until reviewed.

Use these starting points:

• learning-loop/candidates/candidate-template.md
• learning-loop/candidates/example-company-context-update.md

Step 9: Validate, commit, and push

markster-os validate .
markster-os commit -m "docs(context): update company context"
markster-os push

Step 10: Back up or share the workspace

markster-os backup-workspace ~/.markster-os/workspaces/your-company
markster-os export-workspace ~/.markster-os/workspaces/your-company

By default, export-workspace excludes learning-loop/inbox/ so teams can share a cleaner copy without raw notes.

Very Simple Open-Source Workflow

If you are using Markster OS for your company, do this:

1. install the CLI
2. create a workspace in a new Git repo
3. run your AI tool from inside that workspace
4. keep your real company context in company-context/
5. keep raw notes in learning-loop/inbox/
6. keep the local validation hooks installed so validation runs before commits and pushes
7. commit and push approved changes like a normal repo

If you want to hand the setup to Claude, Codex, or Gemini, copy the prompt in setup-prompts/install-workspace.md.

Segments

Before running any playbook, find your business type. Each segment file maps which GOD Engine bricks matter most at your stage, what your key metrics are, and where founders in your category most commonly get stuck.

Full index: playbooks/segments/README.md

Playbooks

Skills

Markster OS currently ships two layers of skills:

• 7 default-installed skills: markster-os plus the 6 core playbook skills
• an extended public skill library in skills/ for more specialized writing, strategy, prep, and review tasks

/markster-os     -> diagnostic operator: Hormozi diagnostic, constraint routing, context pre-fill
/cold-email     -> cold email playbook
/events         -> events playbook
/content        -> content playbook
/sales          -> sales playbook
/fundraising    -> fundraising playbook
/research       -> research prompt library

If you need a specific public skill that is not installed yet:

markster-os list-skills
markster-os install-skills --skill website-copywriter

The repo also includes specialist skills such as:

• strategy: business-advisor, marketing-strategist, sales-strategist, product-owner, startup-coach
• writing: website-copywriter, blog-post-writer, cold-email-copywriter, direct-response, linkedin-post, case-study-builder, partnership-pitch, vc-comms
• prep and review: debrief, event-prep, event-strategist, follow-up, funnel-builder, prospect-brief, vc-review, youtube-watcher
• style references: hormozi, karpathy

See skills/README.md for individual install instructions.

The recommended flow is:

1. create a Markster OS workspace with markster-os init --git --path ...
2. attach it to the company GitHub repo with markster-os attach-remote
3. install slash-command skills with markster-os install-skills
4. install any extra public skills you need with markster-os install-skills --skill <name>
5. run markster-os start inside the workspace to see the readiness checklist
6. run your AI tool from inside the workspace so the skills can resolve the local methodology, playbooks, company context, and learning loop files
7. validate, commit, and push workspace changes through the company repo

Integrations

• ClickUp - task routing and project sync
• More in integrations/README.md

Company Context

The company-context/ folder is the canonical source of truth for how your business should be described across content, outreach, sales, and website copy.

Start here if you want Markster OS to sound like your business instead of generic AI output.

Learning Loop

The learning-loop/ folder is the promotion system for business knowledge.

It separates:

• raw inbox material
• structured candidate updates
• approved canon

This lets you improve the system over time without letting an LLM rewrite the business context arbitrarily.

Workspace Model

The installed skills are lightweight entry points.

The actual working system lives in a Markster OS workspace, usually under:

• ~/.markster-os/workspaces/<slug>/

Run your AI tool from inside that workspace so the skills can read:

• methodology
• playbooks
• company context
• learning loop
• validation rules

The upstream repo is the distribution and template source.
The workspace is the customer-owned operating system.

Backup And Sharing

Recommended default:

• use markster-os backup-workspace for private backups
• use markster-os export-workspace for a shareable copy

export-workspace excludes learning-loop/inbox/ by default so raw notes and transcripts do not get shared accidentally.

For teams, the stronger default is to keep the workspace in its own Git repo and use backup/export as secondary safety nets.

Collaboration

Recommended team model:

• one workspace repo per company
• employees collaborate in that repo
• raw inbox material stays out of Git by default
• approved canon changes are committed and pushed
• pre-commit and pre-push run markster-os validate . locally
• commit-msg enforces type(scope): summary
• use markster-os sync to fetch and pull --rebase before working
• use markster-os commit and markster-os push if you want the CLI to handle the common Git steps

Validation

Markster OS now includes a hard-gate validator for the company context and learning loop structure.

GitHub Actions should fail if:

• required files are missing
• front matter is broken
• headings drift from the expected template
• unauthorized files appear in protected folders
• canon contains prohibited raw or unapproved content patterns

License

Skills, playbooks, templates, and research prompts: MIT License.

ScaleOS is a trademark of Markster. The methodology is open to use under the terms in LICENSE. Not free to rebrand or resell as your own methodology.

Built by Markster. Questions: [email protected]

GitHub: github.com/markster-public/markster-os