Generate Ralph-compatible prompts for entire projects from scratch. Creates comprehensive prompts with architecture phase, implementation phases, testing, and documentation. Use when building complete applications, libraries, CLI tools, or any greenfield project requiring end-to-end development.
/plugin marketplace add adaptationio/Skrillz/plugin install skrillz@skrillzThis skill inherits all available tools. When active, it can use any tool Claude has access to.
Generates comprehensive prompts for building entire projects from scratch using the Ralph Wiggum technique. These prompts structure development into architectural phases (design, implementation, testing, documentation) with clear milestones and incremental completion.
Best For:
Real-World Success:
Ralph Philosophy: These successes work because Ralph embraces that failures are deterministic and fixable. Each iteration learns from the previous. Don't fear failures—they're expected and provide data for improvement through prompt tuning, not tool changes.
Input Required:
Generate prompt with:
Generate a Ralph project prompt for:
Project: [Project description]
Tech stack: [Language, framework, tools]
Features: [List of required features]
Promise: [COMPLETION_PHRASE]
Project Definition Template:
PROJECT: [Project Name]
PURPOSE: [One sentence explaining what it does]
TECH STACK: [Languages, frameworks, tools]
TARGET USERS: [Who will use this]
CORE FEATURES:
1. [Feature 1]
2. [Feature 2]
3. [Feature 3]
NON-GOALS (Out of scope):
- [What this project will NOT do]
- [Explicit exclusions]
SUCCESS DEFINITION:
[Measurable, verifiable completion criteria]
Standard project phases:
| Phase | Name | Purpose | Typical % |
|---|---|---|---|
| 0 | Setup | Project scaffolding, dependencies | 5% |
| 1 | Architecture | Design patterns, structure | 10% |
| 2 | Core | Main functionality | 40% |
| 3 | Features | Additional features | 25% |
| 4 | Testing | Test coverage | 10% |
| 5 | Polish | Documentation, cleanup | 10% |
Each phase needs concrete milestones:
Good Milestones:
Examples:
Use this template:
# Project: [Project Name]
## Vision
[2-3 sentences describing what this project is and why it exists]
## Technical Specifications
- **Language**: [e.g., TypeScript, Python, Go]
- **Framework**: [e.g., Express, FastAPI, none]
- **Database**: [e.g., PostgreSQL, SQLite, none]
- **Testing**: [e.g., Jest, pytest, go test]
- **Build**: [e.g., npm, poetry, go build]
## Core Features
1. [Feature 1]: [Brief description]
2. [Feature 2]: [Brief description]
3. [Feature 3]: [Brief description]
[...]
## Non-Goals
- [Explicit exclusion 1]
- [Explicit exclusion 2]
---
## Phase 0: Project Setup (Foundation)
### Objective
Initialize project structure with all dependencies and configuration.
### Tasks
1. Initialize project structure
2. Set up package management
3. Configure development tools (linting, formatting)
4. Create initial directory structure
5. Add basic configuration files
### Deliverables
- [ ] Project directory created
- [ ] Dependencies installed
- [ ] Linting configured and passing
- [ ] Basic structure in place
- [ ] README with setup instructions
### Verification
```bash
# Project builds
[build command]
# Lint passes
[lint command]
# Basic structure exists
ls -la [expected directories]
PHASE 0 COMPLETE:
- Project initialized: [path]
- Dependencies: Installed
- Lint: Passing
→ Continue to Phase 1
Establish core architecture, interfaces, and patterns.
# Types compile
[typecheck command]
# Config loads
[test config command]
PHASE 1 COMPLETE:
- Architecture: [pattern used]
- Types: Defined and compiling
- Config: Loading correctly
→ Continue to Phase 2
Implement the primary functionality that defines this project.
[List core implementation tasks based on project type]
# Core functionality works
[verification command]
# Tests pass
[test command]
PHASE 2 COMPLETE:
- Core feature 1: Working
- Core feature 2: Working
- Tests: Passing
→ Continue to Phase 3
Add all required features beyond the core.
[List additional feature tasks]
# Features work
[feature verification commands]
# All tests pass
[full test command]
PHASE 3 COMPLETE:
- Feature 1: Working
- Feature 2: Working
- All tests: Passing
→ Continue to Phase 4
Ensure comprehensive test coverage and code quality.
# Coverage report
[coverage command]
# All quality checks
[lint command] && [typecheck command] && [test command]
PHASE 4 COMPLETE:
- Coverage: [X]%
- All tests: Passing
- Quality: All checks green
→ Continue to Phase 5
Complete documentation and final polish.
# Check for TODOs
grep -r "TODO" src/
# Lint final
[lint command]
# Build final
[build command]
# Clean build
[clean and build command]
# All tests
[full test command]
# Lint
[lint command]
# Type check (if applicable)
[typecheck command]
# Manual smoke test
[manual verification steps]
When ALL of the following are true:
Output: <promise>[COMPLETION_PHRASE]</promise>
Update after each phase:
PROJECT PROGRESS:
Phase 0 (Setup): [COMPLETE/IN PROGRESS]
Phase 1 (Architecture): [COMPLETE/IN PROGRESS/NOT STARTED]
Phase 2 (Core): [COMPLETE/IN PROGRESS/NOT STARTED]
Phase 3 (Features): [COMPLETE/IN PROGRESS/NOT STARTED]
Phase 4 (Testing): [COMPLETE/IN PROGRESS/NOT STARTED]
Phase 5 (Documentation): [COMPLETE/IN PROGRESS/NOT STARTED]
Current Phase: [X]
Overall: [description of state]
After multiple iterations on same issue:
## Example: REST API Project
**Input:**
- Project: Todo List REST API
- Tech: Node.js, Express, TypeScript, SQLite
- Features: CRUD todos, user auth, categories, due dates
- Promise: TODO_API_COMPLETE
**Generated Prompt:**
```markdown
# Project: Todo List REST API
## Vision
A RESTful API for managing todo items with user authentication, categories, and due date tracking. Built with TypeScript and Express for type safety and ease of development.
## Technical Specifications
- **Language**: TypeScript
- **Framework**: Express.js
- **Database**: SQLite with better-sqlite3
- **Testing**: Jest with supertest
- **Build**: npm with tsc
## Core Features
1. **User Authentication**: JWT-based auth with signup/login/logout
2. **Todo CRUD**: Create, read, update, delete todo items
3. **Categories**: Organize todos by category
4. **Due Dates**: Set and track due dates with overdue detection
5. **Filtering**: Filter todos by status, category, due date
## Non-Goals
- Frontend/UI (API only)
- Real-time updates (polling only)
- File attachments
- Collaboration/sharing
---
## Phase 0: Project Setup
### Tasks
1. `npm init` with TypeScript configuration
2. Install dependencies: express, better-sqlite3, jsonwebtoken, bcrypt
3. Install dev dependencies: typescript, jest, supertest, eslint
4. Configure tsconfig.json, eslint, jest
5. Create directory structure: src/{routes,controllers,models,middleware}
### Deliverables
- [ ] package.json with all dependencies
- [ ] tsconfig.json configured
- [ ] ESLint configured and passing
- [ ] Directory structure created
- [ ] `npm run build` works
### Verification
```bash
npm run build
npm run lint
ls src/
PHASE 0 COMPLETE: Project scaffolded, builds successfully
npm run build
npm test -- --grep "database"
# Signup
curl -X POST -H "Content-Type: application/json" \
-d '{"email":"test@test.com","password":"password123"}' \
http://localhost:3000/auth/signup
# Login and use token for todos
TOKEN=$(curl -s -X POST -H "Content-Type: application/json" \
-d '{"email":"test@test.com","password":"password123"}' \
http://localhost:3000/auth/login | jq -r '.token')
curl -H "Authorization: Bearer $TOKEN" http://localhost:3000/todos
npm test -- --coverage
npm run lint
npm run build
npm test -- --coverage
npm start # Manual smoke test
When all phases verified and:
Output: <promise>TODO_API_COMPLETE</promise>
## Best Practices
### Project Scope
- Define clear boundaries (non-goals)
- Start smaller than you think necessary
- Add features incrementally
### Phase Management
- Complete phases in order
- Don't skip testing phase
- Document checkpoints
### Verification
- Include actual commands to run
- Make success criteria binary
- Automate where possible
### DO:
- Define clear project scope
- Use all 6 phases
- Include verification commands
- Set realistic feature scope
- Document progress at checkpoints
### DON'T:
- Skip architectural phase
- Rush to implementation
- Leave testing for "later"
- Scope creep mid-project
- Output promise before documentation
## Integration with Ralph Loop
```bash
/ralph-wiggum:ralph-loop "[paste generated prompt]" --completion-promise "YOUR_PROMISE" --max-iterations 100
Recommended iterations by project size:
--max-iterations 60-80--max-iterations 100-150--max-iterations 150-200Tip: For large projects, consider splitting into multiple Ralph sessions:
For single-task prompts, see ralph-prompt-single-task.
For multi-task prompts, see ralph-prompt-multi-task.
For research/analysis prompts, see ralph-prompt-research.
Activates when the user asks about Agent Skills, wants to find reusable AI capabilities, needs to install skills, or mentions skills for Claude. Use for discovering, retrieving, and installing skills.
Activates when the user asks about AI prompts, needs prompt templates, wants to search for prompts, or mentions prompts.chat. Use for discovering, retrieving, and improving prompts.
This skill should be used when the user asks to "create an agent", "add an agent", "write a subagent", "agent frontmatter", "when to use description", "agent examples", "agent tools", "agent colors", "autonomous agent", or needs guidance on agent structure, system prompts, triggering conditions, or agent development best practices for Claude Code plugins.