HarnessCommunity★ 120k

gstack

Updated Jul 5, 2026

Garry Tan's personal Claude Code setup, open-sourced. 23+ slash-command skills covering the full sprint: /office-hours (6 forcing questions), /plan-ceo-review, /plan-eng-review, /design-shotgun, /review, /qa (real browser), /cso (OWASP + STRIDE), /ship, and more. Works across Claude Code, Codex, Cursor, OpenCode, and Factory.

View on GitHub

What it does

/office-hours
Six forcing questions that reframe your product before you write code — challenges premises and extracts the real user pain.
/review
Staff Engineer skill: finds bugs that pass CI but blow up in production; auto-fixes obvious ones, flags blockers.
/qa
Opens a real Chromium browser, clicks through your app flows, finds and fixes bugs, and generates regression tests.
/cso
OWASP Top 10 + STRIDE threat model audit with 17 false-positive exclusions and independent finding verification.
/ship
Sync main, run tests, audit coverage, push, and open PR — bootstraps test frameworks if the project has none.
/autoplan
One command runs CEO → design → eng review automatically and surfaces only the taste decisions that need your approval.

Files (1)

OVERVIEW.mdprimary · markdown · 2.6 KB

## What it does

gstack is Garry Tan's open-source software factory — the setup he uses to ship 3 production services and 40+ features in 60 days while running Y Combinator full time. 23+ slash-command skills organized as a sprint process: Think → Plan → Build → Review → Test → Ship → Reflect. Each skill feeds into the next.

**The sprint skills:**

| Skill | Your specialist | What they do |
|-------|----------------|--------------|
| /office-hours | YC Office Hours | Six forcing questions that reframe your product before you write code. Challenges premises, extracts capabilities you didn't realize you were describing. |
| /plan-ceo-review | CEO / Founder | Rethink the problem. Find the 10-star product. Four modes: Expansion, Selective Expansion, Hold Scope, Reduction. |
| /plan-eng-review | Eng Manager | ASCII diagrams, data flow, state machines, edge cases, test matrix, failure modes. |
| /plan-design-review | Senior Designer | Rates each design dimension 0-10, explains what a 10 looks like, then edits the plan to get there. AI slop detection. |
| /review | Staff Engineer | Find bugs that pass CI but blow up in production. Auto-fixes obvious ones, flags the rest. |
| /investigate | Debugger | Iron Law: no fixes without investigation. Traces data flow, tests hypotheses, stops after 3 failed fixes. |
| /qa | QA Lead | Opens a real Chromium browser, clicks through flows, finds and fixes bugs, generates regression tests. |
| /cso | Chief Security Officer | OWASP Top 10 + STRIDE threat model. 17 false-positive exclusions, 8/10+ confidence gate, independent finding verification. |
| /ship | Release Engineer | Sync main, run tests, audit coverage, push, open PR. Bootstraps test frameworks if you don't have one. |
| /autoplan | Review Pipeline | One command: CEO → design → eng review automatically. Surfaces only taste decisions for your approval. |
| /design-shotgun | Design Explorer | Generates 4-6 mockup variants, opens a comparison board in your browser, iterates. |
| /retro | Eng Manager | Per-person breakdowns, shipping streaks, test health trends. /retro global runs across all projects and AI tools. |

## Install (30 seconds)

Paste in Claude Code:

```
Install gstack: run git clone --single-branch --depth 1 https://github.com/garrytan/gstack.git ~/.claude/skills/gstack && cd ~/.claude/skills/gstack && ./setup
```

Then follow prompts. Works on Claude Code, Codex, OpenCode, Cursor, Factory, Kiro, and Hermes.

## When to use

When you want a structured sprint process — not just individual tools. The skills run in sprint order: each one feeds into the next, nothing falls through the cracks.