🤖 Droidz - Spec-Driven Development for AI

Transform how you build software with AI agents. Droidz is a complete framework that brings structure, standards, and systematic workflows to AI-assisted development.

Instead of ad-hoc prompting and context loss, Droidz gives you:

📋 Spec-driven workflow - Plan → Spec → Tasks → Implementation
🚀 Parallel execution - Run multiple task groups simultaneously (NEW!)
🤖 8 specialized agents - Each expert in their domain
📚 50+ production skills - From TDD to Next.js to security patterns
📏 Team standards - Consistent code across all features
🔄 Repeatable process - Same quality, every time

Stop reinventing workflows. Start shipping faster.

💬 Join Our Discord Community

Built specifically for Ray Fernando's Discord members! 🎯

Get early access, share tips, connect with contributors, and influence future development.

→ Join Discord Community

💝 Support This Project

If Droidz saves you time, consider supporting its development!

→ Donate via PayPal (@gideonapp)

Your support helps maintain and improve this framework! 🙏

With Gratitude

We're deeply thankful for the generosity of friends who keep this framework alive:

Sorennza
Ray Fernando
Douwe de Vries

Every contribution—large or small—directly fuels new features, improvements, and continued open distribution. Thank you for believing in Droidz.

Quick Start

# 1. Navigate to your project
cd /path/to/your/project

# 2. Install Droidz (one command!)
bash <(curl -fsSL https://raw.githubusercontent.com/korallis/Droidz/main/install.sh)

# 3. Follow the interactive menu to choose your AI platform
# (Factory AI, Claude Code, Cursor, etc.)

# 4. Restart your AI tool

# 5. Start using workflows!
> /standards-shaper  # Set up your project standards
> /plan-product      # Plan your product
> /shape-spec        # Design a feature

Why Droidz?
The Droidz Workflow
Installation
Commands Reference
Specialized Agents
Best Practices
Customization
Examples
Troubleshooting

Why Droidz?

The Problem with AI Development

Without structure, AI-assisted development leads to:

❌ Inconsistent code quality
❌ Lost context between sessions
❌ Vague requirements → Rework
❌ No team alignment
❌ Hard to onboard new devs

The Droidz Solution

Droidz brings systematic workflows to AI development:

Planning → Specification → Task Breakdown → Implementation
   ↓            ↓               ↓                ↓
Product      Detailed       Actionable     Guided execution
 docs         spec.md        tasks.md      with standards

Result: Predictable, high-quality output every time.

The Droidz Workflow

Droidz follows a proven 8-phase cycle for building features:

┌─────────────────────────────────────────────────────────────┐
│                    DROIDZ WORKFLOW                          │
├─────────────────────────────────────────────────────────────┤
│                                                             │
│  Phase 0: Setup Standards        → /standards-shaper       │
│           ↓                                                 │
│  Phase 1: Product Planning       → /plan-product           │
│           ↓                                                 │
│  Phase 2: Spec Shaping           → /shape-spec             │
│           ↓                                                 │
│  Phase 3: Spec Writing           → /write-spec             │
│           ↓                                                 │
│  Phase 4: Task Creation          → /create-tasks           │
│           ↓                                                 │
│  Phase 5: Task Orchestration     → /orchestrate-tasks      │
│           ↓                                                 │
│  Phase 6: Implementation         → /implement-tasks  🚀     │
│           │                        (3 MODES!)               │
│           │                                                 │
│           ├─→ A) Parallel (FAST) - All tasks run together  │
│           ├─→ B) Interactive - Live progress updates       │
│           └─→ C) Sequential - One at a time                │
│           ↓                                                 │
│  Phase 7: Continuous Improvement → iterate & refine        │
│                                                             │
└─────────────────────────────────────────────────────────────┘

Why This Works

Context Persists - Specs and standards live in your repo
Clear Handoffs - Each phase produces artifacts for the next
Parallel Execution - Implement multiple task groups simultaneously (NEW!)
Progress Tracking - See real-time updates as tasks complete
Team Alignment - Everyone works from the same spec
Quality Gates - Standards enforced automatically
Scalable - Same process for small features or large systems

🚀 NEW: Parallel Execution

Speed up implementation by 4x! Droidz now supports parallel execution using Factory AI's Droid Exec.

Three Execution Modes

When you run /implement-tasks, choose how to execute:

A) Parallel Execution (FAST) ⚡

All task groups run simultaneously
Uses Factory's headless Droid Exec mode
Auto-loads API key from droidz/config.yml
Bounded concurrency (configurable, default: 4)
Robust error handling
Best for: Multi-group implementations

# Just run the script - API key loads from droidz/config.yml
$ bash droidz/specs/[spec]/implementation/run-parallel.sh

🚀 Starting parallel implementation...
📄 Loading configuration from droidz/config.yml
✅ Using API key from config.yml
⚙️  Autonomy level: medium
🔢 Max parallel: 4

▶️  Starting: 1-database-setup.md
▶️  Starting: 2-api-endpoints.md
▶️  Starting: 3-frontend-ui.md
▶️  Starting: 4-testing.md
✅ Completed: 1-database-setup.md
✅ Completed: 2-api-endpoints.md
✅ Completed: 3-frontend-ui.md
✅ Completed: 4-testing.md

🎉 All task groups completed!

B) Interactive with Live Progress 📊

Sequential execution with real-time TodoWrite updates
See tool calls and results as they happen
Best for: Learning, debugging, following along

C) Sequential Delegation 🔄

Traditional one-at-a-time execution
Simple and reliable
Best for: Single task groups

Requirements for Parallel Mode

One-Time Setup:

# 1. Copy the config template
cp droidz/config.yml.template droidz/config.yml

# 2. Get your Factory API key
# Visit: https://app.factory.ai/settings/api-keys

# 3. Edit droidz/config.yml and add your key:
factory_api_key: "fk-your-key-here"

# That's it! The script will auto-load your key from config.yml
# ⚠️  config.yml is in .gitignore - your key won't be committed

Optional Configuration:

You can also configure in droidz/config.yml:

default_autonomy_level: "low", "medium", or "high" (default: "medium")
max_parallel_executions: 1-10 (default: 4)

Running Parallel Execution:

# Just run the script - API key loads automatically
bash droidz/specs/[your-spec]/implementation/run-parallel.sh

Installation

Prerequisites

One of these AI tools:
- Factory AI
- Claude Code
- Cursor
- Cline
- Codex CLI
- VS Code with AI extension

Install Command

# Navigate to your project first!
cd /path/to/your/project

# Run the installer
bash <(curl -fsSL https://raw.githubusercontent.com/korallis/Droidz/main/install.sh)

What Gets Installed

The installer creates these directories in your project:

your-project/
├── .factory/          # Factory AI droids & commands (if selected)
│   ├── droids/        # 8 specialized droids
│   └── commands/      # 20+ workflow commands
├── .claude/           # Claude Code agents & commands (if selected)
│   ├── agents/        # 8 specialized agents
│   └── commands/      # 20+ workflow commands
└── droidz/            # Shared across all platforms
    ├── standards/     # Your team's coding standards
    │   ├── global/    # Global conventions
    │   ├── backend/   # API, database patterns
    │   ├── frontend/  # Components, styling
    │   └── testing/   # Test patterns
    ├── product/       # Product documentation
    │   ├── mission.md
    │   ├── roadmap.md
    │   └── tech-stack.md
    └── specs/         # Feature specifications
        └── [feature]/
            ├── spec.md
            ├── tasks.md
            └── planning/

Everything is checked into git and shared with your team.

Post-Installation

Restart your AI tool - Required for new commands/agents to load

Verify installation:

> /commands     # See available commands
> /droids       # See available droids (Factory AI)
> /agents       # See available agents (Claude Code)

Commands Reference

Droidz provides 20+ commands organized by phase. Here are the essential ones:

Setup & Planning Commands

`/standards-shaper` ⭐ START HERE

Purpose: Create project-wide coding standards

When to use: Once per project, before any development

What it does:

Detects your tech stack (React, Next.js, TypeScript, etc.)
Creates standards for global, backend, frontend, testing
Generates dos/don'ts for consistency

Output:

droidz/standards/
├── global/
│   ├── coding-principles.md
│   ├── error-handling.md
│   └── security.md
├── backend/
│   ├── api-design.md
│   └── database.md
├── frontend/
│   ├── components.md
│   └── styling.md
└── testing/
    └── testing-patterns.md

Example:

> /standards-shaper

AI: I'll analyze your project and create standards...
AI: Detected: Next.js 14, TypeScript, Tailwind, Prisma
AI: Creating standards with framework-specific examples...

✅ Created 12 standards files
✅ All standards include your actual tech stack

`/plan-product`

Purpose: Define product vision and roadmap

When to use: Starting new products, major milestones, team onboarding

What it does:

Asks questions about your product
Creates mission statement
Builds phased roadmap
Documents tech stack decisions

Output:

droidz/product/
├── mission.md        # Vision, goals, success metrics
├── roadmap.md        # Phases, features, timeline
└── tech-stack.md     # Technologies and rationale

Example:

> /plan-product

AI: What problem does your product solve?
You: Task management for remote teams

AI: Who are your target users?
You: Remote engineering teams of 5-50 people

[More questions...]

✅ Created mission.md
✅ Created roadmap.md (4 phases)
✅ Created tech-stack.md

Specification Commands

`/shape-spec`

Purpose: Shape requirements for a new feature

When to use: Before building anything new

What it does:

Asks clarifying questions
Researches similar implementations (if Exa/Ref available)
Saves requirements and decisions
Creates spec folder structure

Output:

droidz/specs/2024-11-24-user-authentication/
├── planning/
│   ├── requirements.md   # What you need
│   ├── decisions.md      # Key choices made
│   └── visuals/          # Wireframes, screenshots
└── README.md             # Overview

Example:

> /shape-spec

AI: What feature are you planning?
You: User authentication with OAuth

AI: Which OAuth providers?
You: Google and GitHub

AI: What user data do you need to store?
You: Email, name, profile picture

[More questions...]

✅ Saved requirements.md
✅ Saved decisions.md
✅ Created spec folder

`/write-spec`

Purpose: Create detailed specification from shaped requirements

When to use: After shaping (Phase 2 complete)

What it does:

Reads shaped requirements
Generates comprehensive spec document
Includes architecture, APIs, data models, UI, security, testing

Output:

droidz/specs/2024-11-24-user-authentication/
├── spec.md              # ⭐ THE SPEC
├── planning/
│   └── requirements.md
└── README.md

spec.md structure:

# User Authentication Specification

## 1. Overview
## 2. User Stories
## 3. Technical Architecture
## 4. Data Models
## 5. API Endpoints
## 6. UI/UX Flow
## 7. Security Considerations
## 8. Testing Strategy
## 9. Success Metrics

Example:

> /write-spec

AI: Reading shaped requirements...
AI: Generating specification...

✅ Created spec.md (2,400 lines)
✅ Includes:
   - 5 user stories
   - Database schema
   - 8 API endpoints
   - Security checklist
   - 12 test scenarios

Implementation Commands

`/create-tasks`

Purpose: Break spec into implementable tasks

When to use: After spec is written

What it does:

Reads spec.md
Breaks into logical task groups
Creates tasks with dependencies
Adds acceptance criteria

Output:

droidz/specs/2024-11-24-user-authentication/
├── spec.md
├── tasks.md             # ⭐ IMPLEMENTATION TASKS
└── planning/

tasks.md structure:

# Implementation Tasks

## Task Group 1: Database Setup
- [ ] 1.1: Create users table
- [ ] 1.2: Create oauth_tokens table
- [ ] 1.3: Add database indexes

## Task Group 2: OAuth Integration
- [ ] 2.1: Set up OAuth configs
- [ ] 2.2: Create callback endpoints
- [ ] 2.3: Handle token exchange

## Task Group 3: Frontend
- [ ] 3.1: Create login page
- [ ] 3.2: Add OAuth buttons
- [ ] 3.3: Handle redirects

## Task Group 4: Testing
- [ ] 4.1: Unit tests for OAuth
- [ ] 4.2: E2E login flow test
- [ ] 4.3: Security tests

Example:

> /create-tasks

AI: Reading spec.md...
AI: Identifying task groups...
AI: Creating tasks with dependencies...

✅ Created tasks.md
✅ Found 4 task groups
✅ Total: 23 tasks

`/orchestrate-tasks`

Purpose: Plan multi-agent implementation

When to use: Before implementing (optional but recommended)

What it does:

Shows task groups
Lets you assign specialists (if using subagents)
Maps relevant standards to each group
Generates implementation prompts

Output:

droidz/specs/2024-11-24-user-authentication/
├── spec.md
├── tasks.md
├── orchestration.yml    # ⭐ COORDINATION PLAN
└── implementation/
    └── prompts/
        ├── 1-database-setup.md
        ├── 2-oauth-integration.md
        ├── 3-frontend.md
        └── 4-testing.md

Example:

> /orchestrate-tasks

AI: Found 4 task groups. Assign specialists:

1. Database Setup → ?
2. OAuth Integration → ?
3. Frontend → ?
4. Testing → ?

You:
1. backend-specialist
2. backend-specialist
3. frontend-specialist
4. test-specialist

AI: Map standards for each group:

You:
1. backend/*, global/security.md
2. backend/*, global/error-handling.md
3. frontend/*, global/
4. testing/*

✅ Created orchestration.yml
✅ Generated 4 implementation prompts

`/implement-tasks`

Purpose: Execute implementation with parallel execution support

When to use: After tasks are created

Three Execution Modes:

A) Parallel Execution (FAST) 🚀

Uses Factory's Droid Exec (headless mode) for true concurrent processing
Runs all task groups simultaneously
Best for: Multi-group implementations
Requirements: Factory API key in droidz/config.yml (one-time setup)
Features:
- Auto-loads API key from config file
- Bounded concurrency (configurable, default: 4)
- Configurable autonomy level (low/medium/high)
- Robust error handling (if one fails, others continue)
- All droids update same tasks.md file
- Real-time output from all executions

B) Interactive with Live Progress 📊

Uses Task tool + TodoWrite for real-time updates
Sequential execution with live progress tracking visible in session
Best for: Following along, learning, debugging
Features:
- Live TodoWrite updates
- See tool calls and results as they happen
- Track progress as each task group completes

C) Sequential Delegation (SIMPLE) 🔄

Traditional one-at-a-time implementation
Best for: Single task group, simple implementations

Example (Parallel Mode):

> /implement-tasks

How would you like to execute implementation?
A) Parallel Execution (FAST) - Using Droid Exec
B) Interactive with Live Progress - Using TodoWrite
C) Sequential Delegation (SIMPLE)

Enter A, B, or C: A

✅ Created parallel execution setup

PROMPTS: droidz/specs/2024-11-24/implementation/prompts/
├── 1-database-setup.md
├── 2-api-endpoints.md
├── 3-frontend-ui.md
└── 4-testing.md

SCRIPT: run-parallel.sh

TO EXECUTE:
$ bash droidz/specs/2024-11-24/implementation/run-parallel.sh

🚀 Starting parallel implementation...
📄 Loading configuration from droidz/config.yml
✅ Using API key from config.yml
⚙️  Autonomy level: medium
🔢 Max parallel: 4

▶️  Starting: 1-database-setup.md
▶️  Starting: 2-api-endpoints.md
▶️  Starting: 3-frontend-ui.md
▶️  Starting: 4-testing.md
✅ Completed: 1-database-setup.md
✅ Completed: 2-api-endpoints.md
✅ Completed: 3-frontend-ui.md
✅ Completed: 4-testing.md

🎉 All task groups completed!

Utility Commands

`/improve-skills`

Purpose: Enhance AI agent skills

When to use: Customizing agent capabilities

What it does:

Shows current skills
Lets you improve descriptions
Adds examples
Makes skills more discoverable

Specialized Agents

Droidz includes 8 specialized agents, each expert in their domain:

1. product-planner

Expertise: Product strategy, roadmaps, vision

Use when:

Starting new products
Planning major features
Creating product documentation

What it creates:

mission.md - Product vision and goals
roadmap.md - Phased development plan
tech-stack.md - Technology decisions

Best for: Product managers, founders, tech leads

2. spec-shaper

Expertise: Requirements gathering, research

Use when:

Feature requirements are unclear
Need to research implementations
Validating technical approach

What it does:

Asks clarifying questions
Researches patterns (with Exa/Ref)
Documents decisions
Shapes clear requirements

Best for: Early-stage feature design

3. spec-writer

Expertise: Writing detailed specifications

Use when:

Have shaped requirements
Need implementable spec
Building complex features

What it creates:

Comprehensive spec.md with:
- Architecture
- Data models
- API contracts
- UI flows
- Security considerations
- Testing strategy

Best for: Detailed feature planning

4. spec-verifier

Expertise: Specification review

Use when:

Validating specs before implementation
Checking completeness
Finding gaps

What it checks:

Completeness
Clarity
Implementability
Security considerations
Test coverage

Best for: Quality assurance, pre-implementation review

5. tasks-list-creator

Expertise: Breaking specs into tasks

Use when:

Have completed spec
Ready to start implementation
Need task breakdown

What it creates:

tasks.md with:
- Logical task groups
- Clear dependencies
- Acceptance criteria
- Size estimates

Best for: Implementation planning

6. implementer

Expertise: Feature implementation

Use when:

Have tasks defined
Ready to code
Following spec and standards

What it does:

Reads spec and tasks
Implements features
Follows project standards
Writes tests
Updates progress

Best for: Actual development work

7. implementation-verifier

Expertise: Testing and verification

Use when:

Implementation complete
Need to verify against spec
Running test suite

What it does:

Verifies against acceptance criteria
Runs all tests
Checks standards compliance
Creates verification report

Best for: Quality assurance, pre-merge checks

8. spec-initializer

Expertise: Spec folder setup

Use when:

Starting new spec
Need folder structure

What it creates:

Spec folder with proper structure
README.md template
planning/ directory

Best for: Quick spec initialization

Best Practices

✅ DO:

1. Start with Standards

First feature? Run this FIRST:
> /standards-shaper

Why: Standards ensure consistency from day one

2. Always Shape Before Writing

> /shape-spec     # Ask questions, clarify
> /write-spec     # Write detailed spec

Why: Shaping prevents vague specs and rework

3. Break Down Complex Tasks

Don't:
- [ ] Build entire auth system

Do:
- [ ] Create database tables
- [ ] Implement OAuth flow
- [ ] Add login UI
- [ ] Write tests

Why: Small tasks are easier to implement and track

4. Use Orchestration for Big Features

> /orchestrate-tasks   # Plan before implementing

Why: Coordination prevents conflicts and missed requirements

5. Reference Specs in Code

// See droidz/specs/2024-11-24-user-authentication/spec.md
// Section 4.2: OAuth Token Storage
export const saveOAuthToken = async (userId, token) => {
  // ...
}

Why: Links code to requirements

6. Update Standards as You Learn

Found a better pattern? Add to:
droidz/standards/[category]/[pattern].md

Why: Standards evolve with your team

7. Keep Specs Updated

Feature changed during implementation?
Update spec.md to match reality.

Why: Specs are documentation, keep them accurate

8. Use Task Groups

Group 1: Database
Group 2: Backend
Group 3: Frontend
Group 4: Testing

Why: Logical organization, easier to delegate

❌ DON'T:

1. Skip Planning

❌ Jump straight to code
✅ Plan → Spec → Tasks → Implement

2. Write Vague Specs

❌ "Add user authentication"
✅ "Implement OAuth 2.0 with Google/GitHub, 
    JWT tokens, refresh token rotation,
    secure storage per spec section 5.3"

3. Ignore Standards

❌ Implement however you want
✅ Follow droidz/standards/ patterns

4. Mix Concerns in Tasks

❌ - [ ] Build frontend and backend
✅ - [ ] Build backend API
    - [ ] Build frontend UI

5. Lose Context

❌ "What were we building again?"
✅ Read droidz/specs/[feature]/spec.md

6. Skip Testing

❌ Implement without tests
✅ Tests are part of implementation (Task Group 4)

7. Work in Isolation

❌ Solo development without docs
✅ Use workflow so team can collaborate

8. Forget to Commit Droidz Files

❌ .gitignore droidz/
✅ Commit specs, standards, product docs

Examples

Example 1: Building a New Feature

Scenario: Add real-time notifications to your app

# Phase 0: Standards exist (one-time setup)
> /standards-shaper
✅ Standards created

# Phase 1: Product context
> /plan-product
✅ Product docs created

# Phase 2: Shape the feature
> /shape-spec

AI: What feature are you planning?
You: Real-time notifications with WebSockets

AI: What types of notifications?
You: New messages, mentions, system alerts

AI: How should they be delivered?
You: Browser push + in-app toast

[More questions...]

✅ requirements.md created

# Phase 3: Write detailed spec
> /write-spec

✅ spec.md created (1,800 lines)
   Includes: WebSocket architecture, message schema,
   push notification setup, UI components

# Phase 4: Break into tasks
> /create-tasks

✅ tasks.md created
   Found 5 task groups, 28 tasks

# Phase 5: Orchestrate (optional)
> /orchestrate-tasks

You assign:
1. WebSocket Server → backend-specialist
2. Message Schema → backend-specialist
3. Push Notifications → backend-specialist
4. UI Components → frontend-specialist
5. E2E Tests → test-specialist

✅ Implementation prompts generated

# Phase 6: Implement
[Use subagents or implement manually with prompts]

✅ All 28 tasks completed
✅ Tests passing
✅ Feature ready to ship! 🚀

Example 2: Onboarding New Team Member

# New dev joins your team

# 1. They clone the repo
git clone your-repo
cd your-repo

# 2. They install Droidz
bash <(curl -fsSL https://raw.githubusercontent.com/korallis/Droidz/main/install.sh)

# 3. They read your docs
cat droidz/product/mission.md      # Understand the product
cat droidz/product/tech-stack.md   # See tech choices
ls droidz/standards/                # Review coding standards

# 4. They explore past features
ls droidz/specs/                    # See all features built
cat droidz/specs/2024-11-20-*/spec.md  # Read a spec

# 5. They start their first feature
> /shape-spec                       # Learn the workflow
> /write-spec
> /create-tasks
> /implement-tasks

# Result: New dev is productive day 1, following same patterns as team

Example 3: Mid-Project Standards Creation

Scenario: You have a codebase but no standards yet

# Run standards-shaper
> /standards-shaper

AI: I'll analyze your codebase...
AI: Detected:
   - React 18 with hooks
   - TypeScript (strict mode)
   - Tailwind CSS
   - tRPC API
   - Prisma ORM
   - Vitest tests

AI: Creating standards with your patterns...

✅ Created 15 standards files
✅ Examples use YOUR tech stack
✅ Patterns match YOUR code style

# Now all future features follow these standards automatically

Customization

Modifying Standards

Standards are just markdown files. Edit them to match your team:

# Edit any standard
code droidz/standards/frontend/components.md

# Add your own patterns
echo "## Our Custom Button Pattern" >> droidz/standards/frontend/components.md
echo "..." >> droidz/standards/frontend/components.md

# Standards are preserved during Droidz updates

Creating Custom Agents

Add your own specialized agents:

# Factory AI
create .factory/droids/database-expert.md

# Claude Code
create .claude/agents/database-expert.md

Format:

---
name: database-expert
description: PostgreSQL and Prisma expert for database design
color: blue
model: inherit
---

You are a database design expert...

Adding Custom Commands

# Factory AI
create .factory/commands/optimize-queries.md

# Claude Code
create .claude/commands/optimize-queries.md

Troubleshooting

Commands Not Showing Up

Problem: Installed Droidz but commands don't work

Solutions:

Restart your AI tool (required after installation)

Verify installation:

> /commands     # Should show 20+ commands

Check file locations:

ls .factory/commands/  # Factory AI
ls .claude/commands/   # Claude Code

Agents Not Activating

Problem: Agents don't respond or aren't found

Solutions:

Restart your AI tool

Check file structure (must be flat, not nested):

# ✅ Correct
.factory/droids/implementer.md

# ❌ Wrong
.factory/droids/implementation/implementer.md

Verify with:

> /droids      # Factory AI
> /agents      # Claude Code

Standards Not Being Followed

Problem: AI ignores your standards

Solutions:

Reference standards explicitly:

> Use the implementer agent and follow droidz/standards/backend/api-design.md

Check standards exist:
```
ls droidz/standards/
```
Make standards more specific (add examples)

Lost Customizations After Update

Problem: Your standards got overwritten

Solutions:

Check backup folder:
```
ls /tmp/*droidz*backup*/
```

Restore from backup:

cp /tmp/droidz-backup-*/standards/* droidz/standards/

Prevent in future: Commit droidz/ to git

Spec Not Loading

Problem: Agents can't find your spec

Solutions:

Use correct path:

> Read droidz/specs/2024-11-24-feature-name/spec.md

Check file exists:
```
ls droidz/specs/*/spec.md
```
Verify spec is committed to git

Platform Support

Platform	Install Location	Content
Factory AI	`.factory/`	droids/, commands/
Claude Code	`.claude/`	agents/, commands/
Cursor	`.cursor/`	workflows/
Cline	`.cline/`	prompts/
Codex CLI	`.codex/`	playbooks/
VS Code	`.vscode/droidz/`	snippets/
All Platforms	`droidz/`	standards/, product/, specs/

What's Included

8 Specialized Agents

product-planner
spec-shaper
spec-writer
spec-verifier
spec-initializer
tasks-list-creator
implementer
implementation-verifier

20+ Commands

/standards-shaper
/plan-product
/shape-spec
/write-spec
/create-tasks
/orchestrate-tasks
/implement-tasks
/improve-skills
...and more

50+ Production Skills

Core Development: TDD, debugging, security, APIs
Modern Frameworks: Next.js 14, Convex, NeonDB, Tailwind
Tools: Playwright, MCP builder, document processing
Productivity: File organization, continuous improvement

Complete Standards Library

Global conventions
Backend patterns (API, database)
Frontend patterns (components, styling)
Testing strategies

Support & Community

Getting Help

Documentation: Read droidz/standards/RECOMMENDED_WORKFLOW.md after installation
Discord: Join Ray Fernando's community (link at top)
Issues: GitHub Issues

Contributing

Droidz is open source! Contributions welcome:

Improve documentation
Add new skills
Create example workflows
Report bugs

License

MIT License - see LICENSE file for details

Credits

Created by the Droidz community with special thanks to:

Ray Fernando
All our supporters and contributors

Built on principles of spec-driven development adapted for modern AI coding agents.

Ready to transform your AI development workflow?

bash <(curl -fsSL https://raw.githubusercontent.com/korallis/Droidz/main/install.sh)

Questions? Join the Discord! 💬

🤖 Droidz - Spec-Driven Development for AI

💬 Join Our Discord Community

💝 Support This Project

With Gratitude

Quick Start

Table of Contents

Why Droidz?

The Problem with AI Development

The Droidz Solution

The Droidz Workflow

Why This Works

🚀 NEW: Parallel Execution

Three Execution Modes

Requirements for Parallel Mode

Installation

Prerequisites

Install Command

What Gets Installed

Post-Installation

Commands Reference

Setup & Planning Commands

/standards-shaper ⭐ START HERE

/plan-product

Specification Commands

/shape-spec

/write-spec

Implementation Commands

/create-tasks

/orchestrate-tasks

/implement-tasks

Utility Commands

/improve-skills

Specialized Agents

1. product-planner

2. spec-shaper

3. spec-writer

4. spec-verifier

5. tasks-list-creator

6. implementer

7. implementation-verifier

8. spec-initializer

Best Practices

✅ DO:

1. Start with Standards

2. Always Shape Before Writing

3. Break Down Complex Tasks

4. Use Orchestration for Big Features

5. Reference Specs in Code

6. Update Standards as You Learn

7. Keep Specs Updated

8. Use Task Groups

❌ DON'T:

1. Skip Planning

2. Write Vague Specs

3. Ignore Standards

4. Mix Concerns in Tasks

5. Lose Context

6. Skip Testing

7. Work in Isolation

8. Forget to Commit Droidz Files

Examples

Example 1: Building a New Feature

Example 2: Onboarding New Team Member

Example 3: Mid-Project Standards Creation

Customization

Modifying Standards

Creating Custom Agents

Adding Custom Commands

Troubleshooting

Commands Not Showing Up

Agents Not Activating

Standards Not Being Followed

Lost Customizations After Update

Spec Not Loading

Platform Support

What's Included

8 Specialized Agents

20+ Commands

50+ Production Skills

Complete Standards Library

`/standards-shaper` ⭐ START HERE

`/plan-product`

`/shape-spec`

`/write-spec`

`/create-tasks`

`/orchestrate-tasks`

`/implement-tasks`

`/improve-skills`