Complete development workflow suite from Y Combinator CEO Garry Tan's gstack, adapted for OpenClaw/WorkBuddy.

English | 简体中文

• What is gstack-skills?
• What's New in v2.0
• Quick Start
• Available Commands
• Complete Workflow Example
• Installation
• Philosophy
• Documentation
• Contributing
• License

gstack-skills is the OpenClaw/WorkBuddy adaptation of gstack, the open-source development workflow created by Garry Tan, CEO of Y Combinator.

Garry Tan used gstack to write 600,000+ lines of production code in 60 days (35% tests).

This adaptation makes those powerful workflows available to any AI agent running on OpenClaw or WorkBuddy platforms.

• 15 Specialized Tools: Complete suite covering product ideation to deployment
• Automated Workflows: AI executes workflows automatically based on natural language commands
• One-Command Access: Simply say /ship, /review, /qa, etc.
• State Management: Share context between workflow steps
• OpenClaw Native: Built specifically for OpenClaw/WorkBuddy skill system

v2.0 is a complete rewrite that transforms gstack from documentation into fully executable skills:

1. Automated Execution: Skills now execute automatically, not just provide guidance
2. Command Routing: Parse user input and route to appropriate skill
3. State Management: Share data and context between skills
4. Workflow Orchestration: Run complete workflows with single commands
5. Native Integration: Built for OpenClaw/WorkBuddy from the ground up

Just tell OpenClaw/WorkBuddy to install it:

Please install gstack-skills for me 

Or:

Help me install gstack-skills from GitHub: AICreator-Wind/gstack-openclaw-skills 

OpenClaw/WorkBuddy will:

• Clone the repository
• Detect your platform
• Copy skills to the correct location
• Verify installation
• Tell you when it's ready!

That's it! Then restart OpenClaw/WorkBuddy and say /gstack to get started.

For detailed instructions, see INSTALL.md

Fastest method - Run a single script:

git clone https://github.com/AICreator-Wind/gstack-openclaw-skills.git cd gstack-openclaw-skills ./install.sh

git clone https://github.com/AICreator-Wind/gstack-openclaw-skills.git cd gstack-openclaw-skills install.bat

# Clone the repository git clone https://github.com/AICreator-Wind/gstack-openclaw-skills.git cd gstack-openclaw-skills # Copy to OpenClaw skills directory cp -r gstack-skills ~/.openclaw/skills/ # Or for WorkBuddy cp -r gstack-skills ~/.workbuddy/skills/

After installation, restart OpenClaw/WorkBuddy and simply use any gstack command:

# Get help and see all commands User: "/gstack" # Validate an idea User: "/office-hours I have an idea for an AI code reviewer" # Review code User: "/review my current branch" # Deploy User: "/ship the user authentication feature"

User: "/gstack" # AI shows all available commands: # /office-hours - Product ideation and validation # /plan-ceo-review - CEO perspective planning # /plan-eng-review - Engineering architecture review # /review - Pre-merge code review # /qa - Test application and fix bugs # /ship - Automated release workflow # ... (and 9 more)

Once installed, using gstack-skills is simple - just talk to OpenClaw/WorkBuddy!

After installation, restart OpenClaw/WorkBuddy and say:

User: /gstack 

You'll see all available commands.

User: /review AI: Reviewing your current branch... Files changed: 3 Lines added: +127, Lines removed: -45 Critical Issues Found: ❌ 1. SQL Injection Risk in src/db/queries.py:42 Auto-fixes Applied: ✅ Fixed SQL injection ✅ Added rate limiting Status: ⚠️ DONE_WITH_CONCERNS NEXT: Run /qa for testing 

User: /qa AI: Running systematic QA tests... Test Results: • Unit Tests: 42/42 passed ✅ • Integration Tests: 15/15 passed ✅ Health Score: 95/100 Status: ✅ READY TO SHIP 

User: /ship AI: Preparing automated release... 1. Merging base branch... ✅ 2. Running tests... ✅ (57/57 passed) 3. Updating version... ✅ (v2.2.0) 4. Creating PR... ✅ (#247) Status: ✅ DEPLOYED 

You don't need exact commands. Natural language works:

User: review my code AI: Reviewing your current branch... [Same as /review] 

User: deploy to production AI: Preparing automated release... [Same as /ship] 

See CONVERSATION_GUIDE.md for detailed examples and tips.

Here's how you'd use gstack-skills to develop and ship a new feature:

# Step 1: Validate the idea User: "/office-hours I want to add user profiles to my app" AI: Let me validate your startup idea using YC's 6 questions: 1. What is the user's biggest pain point? 2. How does your solution solve it? 3. Why aren't existing solutions good enough? 4. Can you reach these users? 5. Will people pay? 6. How will users discover it? [Guides through all questions] # Step 2: CEO perspective review User: "/plan-ceo-review user profiles" AI: Analyzing from CEO perspective... - Strategic alignment: ✅ - Market opportunity: ✅ - Resource requirements: Medium - Risks: Identified 3 key risks # Step 3: Engineering architecture review User: "/plan-eng-review user profiles architecture" AI: Reviewing technical architecture... - Database design: PostgreSQL with JSONB - API structure: RESTful with GraphQL fallback - Security: JWT + OAuth2 - Scalability: Horizontal scaling ready # Step 4: Implement the feature [Developer writes code...] # Step 5: Code review User: "/review" AI: Reviewing your current branch... Files changed: 12 Lines added: +342 Lines removed: -89 Critical Issues Found: 1. SQL Injection Risk in src/db/queries.py:42 ❌ 2. Missing Authentication in src/api/routes.py:87 ❌ Auto-fixes Applied: 1. ✅ Fixed 2 unused imports 2. ✅ Formatted code with black Status: ⚠️ DONE_WITH_CONCERNS Please fix critical issues before merging. # Step 6: QA testing User: "/qa user profiles" AI: Running systematic QA tests... Test Results: - Health Score: 87/100 - Bugs Found: 5 - Bugs Fixed: 5 - Tests Passed: 42/42 Status: ✅ Ready to ship # Step 7: Deploy User: "/ship" AI: Preparing automated release... 1. ✅ Merged base branch 2. ✅ Ran all tests (42/42 passed) 3. ✅ Updated version to v2.1.0 4. ✅ Generated changelog 5. ✅ Created PR #234 6. ✅ Updated documentation Status: ✅ SHIPPED PR: https://github.com/your/repo/pull/234

Total time: ~15 minutes for end-to-end feature development and deployment

# For OpenClaw cp -r gstack-skills ~/.openclaw/skills/ # For WorkBuddy cp -r gstack-skills ~/.workbuddy/skills/

# Create symbolic link ln -s /path/to/gstack-openclaw-skills/gstack-skills ~/.openclaw/skills/gstack-skills

Copy to your project root:

cp -r gstack-skills/ /path/to/your/project/

# In OpenClaw/WorkBuddy, test installation User: "/gstack" # If installed correctly, you'll see the command overview

"Don't be half-invested, boil the whole lake" - Garry Tan

AI-assisted development should pursue complete implementation, not shortcuts.

• Identify problems, then actually fix them (don't just note them)
• Complete the task (don't leave "todo: optimize" comments)
• 100% quality is achievable with AI assistance (don't settle for "good enough")

When borrowing features from other products, always ask:

1. Why does it work in the original product?
2. Will it succeed or fail in your product?
3. What adaptations are needed for success?

• Demand specific evidence, not vague descriptions
• "10 people said they want it" is worth more than "everyone wants it"
• Focus on actual behavior, not stated interest

• USAGE.md: Complete usage guide with examples
• SKILL.md: Main skill documentation
• CONTRIBUTING.md: How to contribute

gstack-openclaw-skills/ ├── gstack-skills/ # Main skill package │ ├── SKILL.md # Main entry point and router │ ├── office-hours/ # Product ideation │ ├── plan-ceo-review/ # CEO perspective │ ├── plan-eng-review/ # Engineering review │ ├── review/ # Code review │ ├── qa/ # Testing and QA │ ├── ship/ # Deployment │ ├── investigate/ # Debugging │ └── scripts/ # Helper scripts │ ├── command_router.py # Command routing │ └── state_manager.py # State management ├── USAGE.md # Usage guide ├── README.md # This file └── CONTRIBUTING.md # Contributing guide 

When you type a command:

User: "/review my changes"

The command_router.py script:

1. Parses the input: /review + my changes
2. Routes to the appropriate skill: review/SKILL.md
3. Loads the skill instructions
4. AI executes the skill's workflow automatically

Skills can share data through workflow state:

# Workflow starts /state_manager.py init → creates workflow ID: abc12345 # Each skill can read/write state /office-hours → saves validation results /plan-eng-review → reads validation, saves architecture /review → reads architecture, saves code issues /qa → reads code issues, saves test results /ship → reads all state, creates deployment package

Unlike v1.0 (which only provided guidance), v2.0 skills:

1. Analyze context (git status, project structure, etc.)
2. Execute workflows automatically
3. Make decisions (what tests to run, what bugs to fix)
4. Take actions (run tests, fix bugs, create PRs)
5. Report results (comprehensive status reports)

A: The original gstack was built for Claude Code with manual scripts. gstack-skills v2.0 is a complete rewrite for OpenClaw/WorkBuddy that automates execution. You don't need Bun or specific scripts - just use natural language commands.

A: Yes! gstack-skills is platform-agnostic and works with any project.

A: No, gstack-skills doesn't require external dependencies. It uses the tools you already have (git, your test framework, etc.).

A: Each skill provides clear error messages and suggests how to fix issues. Common issues are documented in the USAGE.md troubleshooting section.

A: Yes! Each skill is a separate markdown file that you can customize. See CONTRIBUTING.md for details.

A: /ship is designed to work with existing CI/CD pipelines. It runs tests, generates PRs, and updates documentation - all CI/CD compatible.

We welcome contributions! See CONTRIBUTING.md for details.

Areas where we'd love help:

• Adding more specialized skills
• Improving error handling
• Adding more examples
• Writing tests
• Translating documentation

MIT License - see LICENSE file.

• Garry Tan for creating gstack and sharing the "Boil the Lake" philosophy
• Y Combinator for the office hours framework
• All contributors to gstack-skills

• GitHub: https://github.com/AICreator-Wind/gstack-openclaw-skills
• Original gstack: https://github.com/garrytan/gstack
• OpenClaw: https://openclaw.ai
• WorkBuddy: https://codebuddy.cn

Version: 2.0.0
Last Updated: 2026-03-21
Status: ✅ Production Ready