Awesome Ralph 
A curated list of resources about Ralph (aka Ralph Wiggum), the AI coding technique that runs AI coding agents in automated loops until specifications are fulfilled.
Ralph is a technique created by Geoffrey Huntley for autonomous AI coding. The core concept is elegantly simple:
while :; do cat PROMPT.md | claude-code ; done
Key principles:
- Run an AI agent repeatedly until a PRD/specification is complete
- Progress persists in files and git history, not in context
- Each iteration starts fresh with clean context
- Use "backpressure" (tests, lints, type checks) to validate work
- Philosophy: "Sit on the loop, not in it" — or as Huntley puts it: "deterministically bad in an undeterministic world"
Found this useful? ⭐️ Like this repo to help others discover it
Contents
- Official Resources
- Playbooks & Methodology
- Implementations
- Tutorials & Guides
- Articles & Blog Posts
- Videos & Podcasts
- Community
- Related Tools
- Support
- Contributing
Official Resources
Primary sources from Geoffrey Huntley, the creator of the Ralph technique.
- Ralph Wiggum as a 'Software Engineer' - The definitive introduction. Explains the bash loop, "tuning like a guitar" metaphor, and the economics of autonomous coding.
- Everything is a Ralph Loop - Deeper philosophical exploration covering "reverse mode" clean-rooming, orchestrator patterns, and "The Weaving Loom" concept.
- Don't Waste Your Back Pressure - Technical deep-dive on backpressure — the art of rejecting invalid generations without creating too much resistance.
- I Ran Claude in a Loop for Three Months - Case study: Ralph built CURSED, a Gen Z slang-based programming language with LLVM compiler.
- Too Many Model Context Protocol Servers - Context allocation theory — Ralph minimizes allocation to avoid compaction events.
Playbooks & Methodology
Comprehensive implementation guides covering the 3 Phases, 2 Prompts, 1 Loop workflow.
- How to Ralph Wiggum - Geoffrey Huntley's official playbook covering context management, sandboxing, and prompt patterns.
- ralph-playbook - Comprehensive methodology guide with diagrams, phase explanations, and "signs & gates" steering techniques.
The Core Workflow
Phase 1: Define Requirements — Human + LLM conversation produces JTBD-aligned specifications Phase 2: Planning Mode — Gap analysis generates prioritized TODO list (no implementation) Phase 3: Building Mode — Implement from plan, run tests, commit, repeat
Essential File Structure
project-root/
├── loop.sh # Ralph loop script
├── PROMPT_build.md # Build mode instructions
├── PROMPT_plan.md # Plan mode instructions
├── AGENTS.md # Operational guide (~60 lines max)
├── IMPLEMENTATION_PLAN.md # Prioritized task list (generated)
├── specs/ # Requirement specs (one per JTBD)
└── src/ # Application source code
Implementations
Claude Code Plugins
- ralph-claude-code - Claude Code with intelligent exit detection, rate limiting, circuit breaker, and semantic response analyzer.
- choo-choo-ralph - Beads-powered 5-phase workflow with structured specs, verified implementations, and compounding knowledge—per-project installation with fully customizable workflows.
Standalone Implementations
- ralph-starter - CLI with integrations for GitHub, Linear, and Notion; supports multiple agents, interactive wizard, workflow presets, and cost tracking.
- snartank/ralph - PRD-driven task management with automatic branching, flowchart visualization, and auto-archiving.
- iannuttall/ralph - Minimal file-based agent loop supporting codex/claude/droid/opencode.
- smart-ralph - Spec-driven development workflow. Transforms feature requests into structured specs (research, requirements, design, tasks) then executes them task-by-task autonomously.
- ralph-wiggum-bdd - A standalone Bash script for Behavior-Driven Development with Ralph Wiggum Loop.
- ralph-orchestrator - Rust orchestrator with 7 AI backends, Hat System for specialized personas, and interactive TUI mode.
- nitodeco/ralph - CLI orchestration tool for PRD-driven development with sequential task execution and interactive terminal UI.
- pentoai/ml-ralph - Autonomous ML agent for running experiments using Claude or Codex.
- oh-my-ralph - Opinionated Ralph Wiggum loop implemented in Python.
Tool-Specific Implementations
- ralph-wiggum-cursor - Cursor implementation with token tracking, context rotation at 80k tokens, and interactive setup.
- opencode-ralph-wiggum - OpenCode implementation with real-time status display, mid-loop context injection, and struggle detection.
- ralph (GitHub Copilot) - VS Code extension with visual Control Panel, Progress Timeline, and Fresh Chat Mode.
- ralph-tui - Terminal UI orchestrator connecting to task trackers with interactive PRD creation.
- Goose Ralph Loop Tutorial - Block's Goose implementation with cross-model review and recipe-based workflows.
Multi-Agent Systems
- ralph-loop-agent - Vercel's TypeScript SDK wrapper with verification callbacks and context summarization.
- multi-agent-ralph-loop - Multi-agent orchestration for complex projects requiring parallel work streams.
Tutorials & Guides
- Getting Started with Ralph - Step-by-step quickstart with Claude Code + Docker by Matt Pocock.
- 11 Tips for AI Coding with Ralph Wiggum - AFK coding, HITL Ralph, iteration caps, and progress tracking.
- The Ralph Wiggum Approach - Stop hook mechanism, troubleshooting, and Q&A.
- The Real Ralph Wiggum Loop - Clarifies original bash loop vs Anthropic plugin philosophy.
- Ralph Wiggum Guide - Community-written comprehensive guide.
- Awesome Claude - Ralph Wiggum - Resource page with additional tips and configuration examples.
- How to Build an Effective Long Running Agent Loop in 7 minutes. - Walks you through the entire process from creating a spec, building and polishing a PRD, to running the agent with frameworks like Claude Code, Codex, and OpenCode.
Articles & Blog Posts
- How Ralph Wiggum Went from The Simpsons to the Biggest Name in AI - VentureBeat coverage on the cultural phenomenon.
- A Brief History of Ralph - Comprehensive timeline including Y Combinator hackathon story.
- Ralph Wiggum and AI Coding Loops - From Springfield to real-world software automation.
- Ralph Wiggum Explained: The Claude Code Loop That Keeps Going - Technical explainer of how and why it works.
- 2026: The Year of the Ralph Loop Agent - Predictions and analysis of the Ralph loop's impact on development.
Videos & Podcasts
Videos
- Ralph Wiggum Deep Dive with Geoffrey Huntley - The definitive video. Live coding demo, history, and comparison of bash-loop vs stop-hook implementations.
- [AI That Works Podcast Episode](https: