fstandhartingerfstandhartingerCommunityยท1 Agent Skills Included

ralph-wiggum

Autonomous spec-driven coding with self-correcting build loops

Runs autonomous coding loops that pick specifications, implement features, run tests, and commit verified changes. Eliminates context window overflow and manual supervision by restarting each iteration with fresh context and shared state on disk. Enforces completion through testable acceptance criteria, retry logic, stuck-spec detection, and optional Telegram progress notifications.
npx skills add fstandhartinger/ralph-wiggum --all -g -y
Available:

Directs the agent to read the project constitution as the single source of truth and to detect whether it is in an autonomous Ralph implementation loop or an interactive chat session.

All Skills in This Repository (1)

Pure Emerald Level Indicators

Frequently Asked Questions

FAQPage Schema
How to install Ralph Wiggum?โ–ผ

Run `npx skills add fstandhartinger/ralph-wiggum --all -g -y` in your terminal to install the skill globally.

What is the Ralph Wiggum loop?โ–ผ

It is an autonomous coding loop where the agent picks one specification, implements it, verifies acceptance criteria, commits, and repeats with a fresh context window each iteration.

How does Ralph Wiggum know a task is done?โ–ผ

The agent only outputs a DONE completion signal when all acceptance criteria pass and changes are committed; otherwise the shell loop retries automatically.

Does Ralph Wiggum work with Codex, Gemini, and Copilot?โ–ผ

Yes. It ships with ready-made loop scripts for Claude Code, OpenAI Codex, Google Gemini, and GitHub Copilot in both Bash and PowerShell.

Do I need detailed plans before running Ralph?โ–ผ

No. Most projects work directly from numbered spec files with clear acceptance criteria; the optional plan mode is only for detailed task breakdowns.

Related Repositories in Software Engineering

View All in Software Engineeringโ†’