Ralph Wiggum Technique

The Ralph Loop (aka Ralph Wiggum technique) is an autonomous AI coding pattern used to build LLM Wiki. Named after Ralph Wiggum from The Simpsons — lovable but forgetful, which mirrors how AI agents behave without persistent memory.

The Core Concept

while :; do cat PROMPT.md | claude-code ; done

A bash loop that repeatedly spawns fresh AI agent instances. Each iteration gets a clean context window. Memory persists only through files on disk — git history, progress files, PRD documents.

Origins

Who	Role
Geoffrey Huntley	Created the pattern (~May 2025). “Sit on the loop, not in it.”
Matt Pocock	Popularized it (Dec 2025). 11 tips, skills repo, Sandcastle library.
Ryan Carson (Snarktank)	Built the most starred implementation (15,861 stars). Runs 3 parallel instances.

How It Works

Phase 1: Define Requirements

Human + LLM produce a PRD (Product Requirements Document). Break work into user stories, each small enough for one context window.

Phase 2: The Ralph Loop

Each iteration:

1. Read prd.json and progress.txt
2. Pick highest priority incomplete story
3. Implement the story
4. Validate (typecheck, lint, test — “backpressure”)
5. Commit with passing checks
6. Update prd.json (mark passes: true) and append learnings to progress.txt
7. Context cleared — next iteration starts fresh

Key Files

File	Purpose
prd.json	User stories with passes: true/false status
progress.txt	Append-only learnings for future iterations
CLAUDE.md	Prompt fed to agent each iteration

Matt Pocock’s Workflow

Idea -> /write-a-prd -> PRD -> /prd-to-issues -> Ralph Loop -> Manual QA

His Key Tips

1. Start hands-on, then go AFK — watch Ralph before letting it run unsupervised
2. Define scope as end state, not implementation steps
3. Use feedback loops — types, tests, linting must pass (“backpressure”)
4. Take small steps — each PRD item fits in one context window
5. Prioritize risky tasks first — architecture before UI polish
6. Explicitly define quality expectations — “Agents amplify what they see”
7. Tune it like a guitar — observe failures, add guardrails reactively

Key Resources

Articles

• Geoffrey Huntley: ghuntley.com/ralph/
• Matt Pocock: aihero.dev/tips-for-ai-coding-with-ralph-wiggum

Repos

• snarktank/ralph (15,861 stars)
• mattpocock/skills (write-a-prd, grill-me, tdd)

Videos

• Matt Pocock overview: youtube.com/watch?v=_IK18goX4X8
• Geoffrey Huntley deep dive: youtube.com/watch?v=SB6cO97tfiY

Relevance to LLM Wiki

The Ralph Loop is used in two ways:

1. Build methodology — LLM Wiki itself was built using Ralph loops (/ralph skill, prd.json, progress.txt)
2. Wiki operations — ingest/lint could run as a Ralph-style loop: process sources one at a time, each in a fresh context, progress tracked in log.md