| name | ralph-init |
| description | Creates AGENTS.md, prd.json, and progress.log to initialize the Ralph environment. Use when setting up a new project for autonomous execution. |
Ralph Init Skill
Load this skill when running /ralph-init. Creates the three required Ralph files.
File 1: AGENTS.md
Universal project context file. Must contain real, runnable commands — no placeholders.
Structure:
# [Project Name]
## Project Overview
[2-3 sentences: what it does, main architecture, primary users]
## Build & Test Commands
- Build: `[exact command]`
- Test: `[exact command]`
- Lint: `[exact command]`
- Typecheck: `[exact command]`
- Dev server: `[exact command]`
## Technology Stack
- Language: [e.g., TypeScript 5.x]
- Framework: [e.g., Next.js 14]
- Database: [e.g., PostgreSQL 16 via Prisma]
- Testing: [e.g., Vitest + Testing Library]
- Styling: [e.g., Tailwind CSS 3.x]
## Project Structure
- `src/` — Source code
- `src/components/` — UI components
- `src/lib/` — Shared utilities
- `tests/` — Test files (mirrors src/ structure)
## Code Style
- [Specific rules, e.g., "Use Zod for all API input validation"]
- [Naming conventions, e.g., "React components in PascalCase, utilities in camelCase"]
- [File conventions]
## Testing Requirements
- Unit tests required for all business logic
- [Coverage threshold if applicable]
- Integration tests for all API endpoints
## Git Workflow
- Branches: `feature/description`, `fix/description`
- Commits: `type(scope): description` (conventional commits)
- No direct commits to `main`
## Security
- API keys via environment variables only — never hardcoded
- [Other project-specific boundaries]
File 2: prd.json
Complete task breakdown. Be exhaustive. Each task should be 15-60 minutes of work.
{
"project": "Project Name",
"description": "What this project builds",
"tasks": [
{
"id": "task-001",
"title": "Specific, actionable task title",
"priority": "high",
"status": "pending",
"dependencies": [],
"acceptance_criteria": [
"Concrete, binary criterion — either it passes or it does not",
"Another specific, testable criterion"
]
}
]
}
Task quality rules:
high priority: core functionality that blocks other tasks
medium priority: features and functionality
low priority: polish, docs, optional improvements
- Acceptance criteria must be binary (pass/fail), never subjective ("looks good")
- Dependencies must reference real task IDs defined in the same file
- Sequence tasks so that infrastructure comes before features
File 3: progress.log
# Ralph Wiggum Progress Log
# Project: [Project Name]
# Started: [YYYY-MM-DD HH:MM:SS]
Completion
After creating all three files, report:
Ralph environment initialized.
AGENTS.md — [N] sections
prd.json — [N] tasks ([N] high, [N] medium, [N] low priority)
progress.log — ready
Run /ralph-loop to start.