What is the Ralph Loop in Wave Orchestration?

The Ralph Loop is an iterative pattern where an agent repeatedly reviews and improves its own work across multiple cycles to achieve higher quality and accuracy than a single-shot prompt.

How does this skill handle multiple tasks at once?

It uses Swarm Orchestration to decompose a large goal into smaller, unblocked sub-tasks, then spawns background agents to execute those tasks in parallel 'waves'.

Where does the skill store its progress?

It maintains state locally in the .claude/ directory, specifically using wave-state.local.md for configuration and the wave-outputs/ folder for individual agent results.

Can I monitor the progress of a wave?

Yes, you can use the /wave-status command to check the current iteration, see which agents are running, and view what work has been completed.

When should I choose Wave Orchestration over standard Claude commands?

Use it for complex, multi-file features or research projects that benefit from parallelization and need multiple rounds of refinement to meet quality standards.

Wave Orchestration

Name: Wave Orchestration
Author: ferni-ai

byferni-ai

0•

Productivity & Workflow

Orchestrates complex development tasks using iterative parallel agent execution and swarm coordination patterns.

Wave Orchestration is a high-level coordination skill for Claude Code designed to handle large-scale features and research tasks that are too complex for a single prompt. By combining the 'Ralph Loop' for iterative self-improvement with 'Swarm Orchestration' for parallel background execution, it breaks down massive goals into manageable waves of concurrent work. The skill maintains a persistent state in your local directory, allowing Claude to autonomously decompose tasks, spawn workers, and synthesize results until specific completion criteria are met.

Key Features

01Automated Synthesis: Intelligently combines results from multiple parallel agents into a cohesive final product.

02Parallel Agent Execution: Launches multiple background workers to process independent tasks simultaneously.

03Iterative Ralph Loop: Implements a cycle of assessment and refinement to ensure high-quality output.

04Background Processing: Allows for complex, long-running workflows that don't block immediate interactions.

05Stateful Task Management: Tracks progress and agent outputs via local markdown files for transparency.

060 GitHub stars

Use Cases

01Executing large-scale codebase refactors that require iterative verification and parallel processing.

02Building multi-component features that require simultaneous backend, frontend, and test updates.

03Conducting deep-dive research or audits across multiple domains with consolidated summary reporting.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add ferni-ai/wave-orchestration wave-orchestration

For use in Claude.ai and ChatGPT

Download Skill