Complete collection of AI agent skills including: - Frontend Development (Vue, React, Next.js, Three.js) - Backend Development (NestJS, FastAPI, Node.js) - Mobile Development (React Native, Expo) - Testing (E2E, frontend, webapp) - DevOps (GitHub Actions, CI/CD) - Marketing (SEO, copywriting, analytics) - Security (binary analysis, vulnerability scanning) - And many more... Synchronized from: https://skills.sh/ Co-Authored-By: Claude <noreply@anthropic.com>
3.3 KiB
3.3 KiB
name, description
| name | description |
|---|---|
| ralph | RalphLoop 'Tackle Until Solved' - Autonomous agent iteration for complex tasks. Use this for architecture, systems, multi-step implementations. Always uses Ralph Orchestrator. |
RalphLoop "Tackle Until Solved" Autonomous Agent
This is an alias for /brainstorming with RALPH_AUTO=true always enabled. Use this for complex tasks that benefit from autonomous iteration until completion.
When to Use
Use /ralph for:
- Architecture and system design
- Multi-step implementations (5+ steps)
- Complex features requiring multiple iterations
- Tasks with multiple phases or dependencies
- Production-quality implementations
- "Tackle until solved" scenarios
What It Does
When you invoke /ralph, it automatically:
- Analyzes task complexity
- Delegates to Ralph Orchestrator for autonomous iteration
- Runs continuous improvement loops until completion
- Presents validated design/implementation
Usage
/ralph "Build a multi-tenant SaaS platform with authentication, billing, and real-time notifications"
How Ralph Works
Ralph runs autonomous iterations:
- Creates task in
.ralph/PROMPT.mdwith success criteria - Iterates continuously until all criteria are met
- Updates progress in
.ralph/state.json - Outputs final result to
.ralph/iterations/final.md
Configuration:
- Max iterations: 100 (configurable via
RALPH_MAX_ITERATIONS) - Max runtime: 4 hours (configurable via
RALPH_MAX_RUNTIME) - Agent: Claude (configurable via
RALPH_AGENT)
Difference from /brainstorming
| Feature | /brainstorming | /ralph |
|---|---|---|
| Simple tasks | Direct mode, fast | Still uses Ralph |
| Complex tasks | Auto-delegates to Ralph | Always uses Ralph |
| User control | Manual opt-in available | Always autonomous |
| Best for | Quick decisions, features | Architecture, systems |
Environment Variables
# Choose agent
RALPH_AGENT=claude|gemini|kiro|q|auto
# Max iterations
RALPH_MAX_ITERATIONS=100
# Max runtime (seconds)
RALPH_MAX_RUNTIME=14400 # 4 hours
# Enable verbose output
RALPH_VERBOSE=true
Files Created
.ralph/
├── PROMPT.md # Task with success criteria
├── ralph.yml # Configuration
├── state.json # Progress tracking
└── iterations/
├── 001.md # First iteration
├── 002.md # Second iteration
└── final.md # Validated result
Monitoring Progress
While Ralph is running:
# Check state
cat .ralph/state.json | jq '.iteration, .status'
# View latest iteration
cat .ralph/iterations/final.md
Examples
Architecture:
/ralph "Design a microservices architecture for an e-commerce platform"
Multi-step feature:
/ralph "Implement authentication, authorization, and user management"
System integration:
/ralph "Integrate Stripe billing with webhooks and subscription management"
Stopping Ralph
Press Ctrl+C to stop. Ralph saves progress and can be resumed later.
Technical Details
- Wrapper:
/home/uroma/obsidian-web-interface/bin/ralphloop - Integration:
/home/uroma/.claude/skills/brainstorming/ralph-integration.py - Ralph Orchestrator: https://github.com/mikeyobrien/ralph-orchestrator