gstack
Use Garry Tan's exact Claude Code setup: 23 opinionated tools that serve as CEO, Designer, Eng Manager, Release Manager, Doc Engineer, and QA
gstack gives Claude Code 23 role-based slash commands, CEO, engineering manager, security officer, release engineer, and more, so AI behaves like a structured virtual team instead of a generic chatbot.
gstack is a collection of opinionated tools designed for use with Claude Code, the AI coding assistant. It treats Claude Code like a virtual engineering team rather than a single chatbot. Instead of writing one prompt and hoping for good output, you invoke specialized roles through slash commands: a CEO who challenges product decisions, an engineering manager who locks down architecture, a designer who catches AI sloppy output, a code reviewer, a quality assurance lead, a security officer who runs audits, and a release engineer who ships pull requests. The package includes 23 such role-based tools plus 8 power tools, all written as Markdown slash commands.
The idea is that a single builder using AI agents can move faster than a traditional team — but only if the AI has structure. Without prompts that put it in a specific role with clear standards, AI tends to produce mediocre work. gstack provides that structure as a free, MIT-licensed system you install once and reuse across projects.
You would use this if you are a founder or technical CEO who still wants to ship code, a first-time Claude Code user looking for structured starting points instead of a blank prompt, or a tech lead who wants rigorous review and release automation on every change. Setup is a single git clone and a setup script, with a team mode that bootstraps a repo so collaborators get the same tools automatically. The package supports several other AI coding agents besides Claude Code through a host detection system.
Where it fits
- Have a CEO agent challenge your product decisions before you build a feature, so you ship the right thing.
- Run a security officer audit on every pull request without writing a security review prompt from scratch.
- Use a release engineer slash command to automate creating pull requests and changelogs.
- Bootstrap a new repo in team mode so all collaborators get the same AI agent roles installed automatically.