ralph-wiggum

Execute spec-driven development loops and emit DONE when acceptance criteria pass.

284|30|Updated Jan 14, 2026
One-click install
npx skills add https://github.com/fstandhartinger/ralph-wiggum --skill ralph-wiggum-fstandhartinger
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ralph-wiggum
Source: https://github.com/fstandhartinger/ralph-wiggum/tree/main/skills/ralph-wiggum
Command: npx skills add https://github.com/fstandhartinger/ralph-wiggum --skill ralph-wiggum-fstandhartinger

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill enables autonomous AI agents to execute spec-driven development loops and declare completion only when acceptance criteria are fully met.

Core Features & Use Cases

  • Fresh context per loop to prevent context drift and ensure reliable iterations.
  • Shared state on disk for specs, plans, and history (e.g., IMPLEMENTATION_PLAN.md, specs/ and ralph_history.txt).
  • Completion verification that requires acceptance criteria, tests, and the DONE signal: <promise>DONE</promise>.
  • Deterministic workflow mirroring Geoffrey Huntley's loop: orient, pick task, implement & test, commit, and output DONE.

Quick Start

Install and set up Ralph Wiggum using skill installers:

  • Using Vercel's add-skill: npx add-skill fstandhartinger/ralph-wiggum
  • Using OpenSkills: openskills install fstandhartinger/ralph-wiggum The agent will guide you through a lightweight setup:
  1. Quick Setup — Create directories, download scripts
  2. Project Interview — Align on vision and goals
  3. Constitution — Establish session-wide rules
  4. Next Steps — Begin creating specs and running loops

Frequently Asked Questions about ralph-wiggum

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate spec-driven development loops with an autonomous AI agent?

Spec-driven development loops work by having an autonomous agent orient, pick a task, implement and test, commit changes, and output a DONE signal. It maintains fresh context per loop and persists specs, plans, and history on disk to prevent context drift.

What's the best way to ensure an autonomous coding agent stops only when acceptance criteria are met?

To ensure an autonomous coding agent stops only when criteria are met, require completion verification that checks acceptance criteria, runs tests, and emits a <promise>DONE</promise> signal. This deterministic workflow prevents premature task completion declarations.

Can I run autonomous software engineering loops using a Bash runtime?

Yes, you can run autonomous software engineering loops using a Bash runtime. The workflow requires Bash and included scripts to load specs, run tests, commit changes, and emit the completion signal across multiple specifications.

How do I prevent context drift during iterative AI software development tasks?

To prevent context drift during iterative AI software development tasks, maintain fresh context per loop. This is achieved by persisting shared state on disk, including implementation plans, specifications, and history files, ensuring each iteration starts with a clean state.

What is the setup process for autonomous spec-driven coding sessions?

The setup process for autonomous spec-driven coding sessions involves creating directories, downloading scripts, conducting a project interview to align on goals, establishing a constitution for session-wide rules, and then creating specs to begin running loops.