Skills

GStack

Open agent skill stack for planning, browsing, QA, shipping, and product workflows.

MIT License
garrytan Maintainer
2026-04-19 Verified
Overview

What is GStack?

GStack is a collection of agent skills and workflows that package repeatable planning, browsing, QA, design, review, and shipping behavior into reusable instructions.

Reusable workflow knowledge

GStack packages processes as skills that agents can invoke repeatedly.

This reduces prompt drift and gives agents better operating habits.

Broad product-building coverage

The stack spans planning, browsing, QA, design, review, and deployment workflows.

A useful agent needs more than code generation; it needs operating procedures.

Open skill format

The repository can be inspected, forked, and adapted.

Teams can learn from the skill structure and tune it to their own projects.
Use cases

What GStack is built for

01

QA and browser checks

Use GStack-style skills to make agents verify web apps with browser evidence.

02

Plan review

Use structured review skills before implementing large product changes.

03

Shipping workflows

Help agents follow a consistent test, commit, push, and deploy flow.

Quick start

Get started in seconds

terminal
$ git clone https://github.com/garrytan/gstack.git
Comparison

How it stacks up

GStack is a process layer

vs single prompt libraries

Instead of collecting prompts, GStack focuses on reusable operational workflows that change how agents behave over a task.

FAQ

Frequently asked questions

What should I check before using GStack?

Evaluate GStack by reading its official source, then running one workflow end to end. Check when the skill should be invoked, what inputs it expects, what evidence it collects, and how easy it is to edit or version.

Is GStack open source?

GStack is listed with MIT based on the official source links in this profile. Re-check the repository, model card, or docs before production use.

Who should evaluate GStack?

GStack is most worth evaluating for builders who want reusable agent workflows instead of one-off prompts.