Company OS Starter Kit workflow

A company knowledge store template for Claude Code: five skills, six commands, rules and a safety hook.

  • 104 ★ on the repository
  • Last commit
  • MIT
Workflow

Extended access

Why this level

  • Hooks run bash scripts before every tool call, and the template rules control what the agent does with client data.
All reasons and checks

workflowsio/company-os-starter-kit

Install

Works in

In your terminal, with SkillFoxx CLI

npx skillfoxx add workflows/company-os-starter-kit

Detects the agents on your machine, shows the access level and pins the version.

Other ways to install

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • Hooks run bash scripts before every tool call, and the template rules control what the agent does with client data.

This is third-party code. Review the repository files before installing.

What it does

Company OS Starter Kit is a ready folder structure you copy into your project so the agent already knows the company at the start of every session. At its center is `CLAUDE.md` pointing to the business description, the skill library, connected systems and rules. Content is split across five roots: `company-wiki`, `clients`, `raw`, `archive` and `.claude`. `.claude` holds five starter skills for sales and marketing: call preparation, GTM strategy, an ideal customer model, LinkedIn posts and cold outreach copy. Next to them are six `/workflows` commands (`brainstorm`, `plan`, `work`, `review`, `compound`, `swarm`), rules for different file types and a `safety-guard.sh` hook that checks every agent action before it runs. The `refresh-counts.mjs` script recomputes numbers in the docs and reports drift in check mode.

Who it is for. Sales and marketing teams building a shared knowledge base for Claude Code agents who want it versioned in git.

Good fit when

  • A team wants the agent to know clients, pricing and brand voice without repeated explanations.
  • Skills have multiplied and need a single flat level so the loader finds them all.
  • A cold email or LinkedIn post is being drafted by the rules from the skill library.
  • Agents touch live systems and every action needs a role-based check first.

Not a fit when

  • You need a single ready skill: take a standalone one, the template unpacks a whole structure.
  • The project is technical, without sales or client material.
  • The team does not use bash: hooks and scripts rely on it.

Example request

Read CLAUDE.md and INDEX.md, help fill in the company block and build an ideal customer model for a new fintech product.

Limitations

This is a starter template, not a package: the structure is copied with `rsync` and adapted to the company. The five skills are written for GTM and other areas are covered by `skills-guide.md`. A skill must sit exactly one level below `.claude/skills`: the loader silently skips nested folders. The safety hook relies on the `.claude/safety-roster` role list and identifies a person by the commit author email, so anyone who edits repository settings can impersonate another. Known gaps are listed in the script header. The refresh script needs Node.js 18 or newer.

How to disable. Delete `.claude/hooks` and the hook entries in `.claude/settings.json`, then the skills and commands in `.claude`. The other folders hold plain markdown files.

FAQ

How do I set the template up in a project?

Clone the repository and copy the tree into your project with `rsync`, excluding `.git` and the root `README.md`. Then create your own README and check the scripts with the quick start commands.

Which skills are included?

Five: call preparation, GTM strategy, ideal customer modelling, LinkedIn posts and outbound copy. Each lives in `.claude/skills/<name>/SKILL.md`.

What does the safety-guard hook do?

It runs before every tool call, reads the whole action and blocks risky ones through an exit code. The decision depends on the person's role: owner, trusted, team or unresolved.

Why are skills kept without nested folders?

The loader finds skills only one level under `.claude/skills`. It silently skips folders nested by service line, so the line is encoded in the name.

What was removed in this version?

The standalone plugin: its commands moved to `.claude/commands/workflows`. Old paths and the reasons for removal are recorded in `CHANGELOG.md`.

Discussion

No questions or reviews yet.

Workflows

A Claude Code workspace template: notes, research, an article in your voice, then LinkedIn, newsletter and social versions.

WorkflowMinimal access223Repository stars

Claude Code skills for founders: pricing and launch plans, PRDs, landing page reviews, social posts and cold outreach

PluginMinimal access677Repository stars

A library of 64 skills for founders: GTM, growth, SEO, email, graphics and video, installed with one npx command for Claude Code, Codex and more

SkillExtended access707Repository stars

AI Marketing Suite

AI Marketing Suite for Claude Code

A set of marketing skills for Claude Code: site audit, copywriting, emails, content calendars, competitor analysis and client reports

SkillExtended access2.9KRepository stars

AI Collab Playbook bundles 19 skills for Claude Code and Codex: paper review, Codex session recovery, anti-AI-tone editing and research-workflow helpers

SkillMinimal access473Repository stars

54 skills for marketing and sales in Claude Code: SEO, content, cold email, launches, ads and CRM

PluginExtended access111Repository stars

README in short

The README describes the five-root structure, a terminal quick start and a Claude-prompt route, the contents of `.claude` and how the safety hook works. Version changes are listed separately.

README badge

Report a problem