Files

uroma 87748afb75 feat: Complete sync of all Claude Code CLI upgrades

- Add all 21 commands (clawd, ralph, prometheus*, dexto*)
- Add all hooks (intelligent-router, clawd-*, prometheus-wrapper, unified-integration-v2)
- Add skills (ralph, prometheus master)
- Add MCP servers (registry.json, manager.sh)
- Add plugins directory with marketplaces
- Add health-check.sh and aliases.sh scripts
- Complete repository synchronization with local ~/.claude/

Total changes: 100+ new files added
All integrations now fully backed up in repository

🤖 Generated with Claude Code
Co-Authored-By: Claude <noreply@anthropic.com>

2026-01-27 20:39:25 +00:00

3.3 KiB

Executable File

Raw Blame History

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.md with 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: ~/.claude/skills/bin/ralphloop
Integration: ~/.claude/skills/brainstorming/ralph-integration.py
Ralph Orchestrator: https://github.com/mikeyobrien/ralph-orchestrator

3.3 KiB Executable File Raw Blame History