---
title: "ralph-wiggum"
description: "Autonomous, iterative, self-referential AI development loop."
canonical: https://agentpluginsdirectory.com/plugins/ralph-wiggum
last-updated: 2026-10-21
---

# ralph-wiggum
Autonomous, iterative, self-referential AI development loop.
- Slug: ralph-wiggum
- Publisher: Burak Kaygusuz
- Repository: https://github.com/burakkaygusuz/ralph-wiggum
- Manifest: plugin.json
- Version: 1.0.0
- License: Apache-2.0
- Category (editorial): agent-tooling
- Skills: 3 (cancel-ralph, ralph-loop-help, ralph-loop)
- MCP servers: 0
- Stars: 0
- Repository created: 2026-08-26
- Repository last pushed: 2026-08-26
- Publisher type: User
- Listing: https://agentpluginsdirectory.com/plugins/ralph-wiggum
- Schema: https://agent-plugins.org/schemas/1.0.0/plugin.schema.json

## What ralph-wiggum does, in the publisher's words

Ralph Wiggum is an autonomous development loop plugin for AI coding assistants, implementing Geoffrey Huntley's Ralph Technique. Instead of stopping after one turn or guessing if the code works, it keeps your agent iterating over a living implementation plan, inspecting test outputs, and self-correcting until objective criteria (compiler, linter, tests) confirm complete success.

- Zero-Babysitting Autonomy: The agent picks the next priority from the implementation plan, writes code, runs tests, and self-corrects without requiring manual intervention every turn.
- Zero Runtime Dependencies: Pure Bash script engine. No Python, Node.js, Ruby, or external pip/npm packages required.
- Living Backlog & Persistent Scheduler: Tracks work items, acceptance criteria, and verified progress in.ralph/fix_plan.md.
- No More Half-Done Code: Strictly terminates only when your success promise ( COMPLETE ) is genuinely verified and all work items are complete.

From the project README, punctuation lightly normalized. Full text: https://raw.githubusercontent.com/burakkaygusuz/ralph-wiggum/HEAD/README.md

## Skills

- cancel-ralph: Cancel and clean up an active Ralph loop session. Use this skill when the user asks to stop, abort, cancel, or terminate a running loop. DO NOT use when starting or querying help for a loop.
- ralph-loop-help: Explain the Ralph Wiggum iterative loop methodology, configuration flags, prompt writing patterns, and available commands. Use this skill when the user asks how Ralph loop works, asks for parameter explanations, or needs help structuring a loop prompt. DO NOT use when the user asks to start or canc…
- ralph-loop: Run an autonomous, iterative development loop using the Ralph Wiggum technique. Use this skill when the user wants continuous test-driven iteration, automated refactoring until tests pass, or execution bounded by a completion promise. DO NOT use for one-shot bug fixes, single-pass queries, or simpl…

Descriptions come from the frontmatter of each SKILL.md, punctuation lightly normalized.