ralph-wiggum
Skill3 starsChanged 3 months agoMITView on GitHub
Scaffold a project for the Ralph Wiggum autonomous control loop. Use when setting up a new project for autonomous iteration, creating loop configuration files, or preparing specs for AI-driven development. Supports single-repo and multi-repo workspaces.
7 other projects use a version of this.
Adopted by
Most recently taken up by
- myths-labs/musetheir own version, first seen Nov 6, 2026
- jkitchin/knowhowtheir own version, first seen Nov 1, 2026
- fstandhartinger/ralph-wiggumtheir own version, first seen Oct 8, 2026
- agrimsingh/ralph-wiggum-cursortheir own version, first seen Oct 8, 2026
- talgacapri/pm-ostheir own version, first seen May 23, 2026
- nbirkner/digital-braintheir own version, first seen Apr 12, 2026
Other versions7 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 7
- agrimsingh/ralph-wiggum-cursor★ 496same name
- fstandhartinger/ralph-wiggum★ 301same name
- jkitchin/knowhow★ 42same name
- myths-labs/muse★ 38same name
- ai4brands-design/claude-skills★ 2same name
- nbirkner/digital-brain★ 0same name
- talgacapri/pm-os★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository41 other files it gives its agents.
- cadence-goalsSkill
- ci-feedback-loopSkill
- compoundSkill
- decision-rights-doctrineSkill
- diataxisSkill
- dispatch-gateSkill
- drift-checkSkill
- family-checkSkill
- fix-then-re-review-ladderSkill
- github-projectsSkill
- gitSkill
- goal-treeSkill
- handoffSkill
- improvement-loopSkill
- kb-captureSkill
- kb-compileSkill
- kb-coreSkill
- kb-lessons-bridgeSkill
- kb-lintSkill
- l1-supervisorSkill
- l2-supervisorSkill
- lessons-learnedSkill
- lN-lifecycle-doctrineSkill
- lN-review-doctrineSkill
- medical-recordsSkill
- operator-interview-doctrineSkill
- operator-interviewSkill
- operator-reviewSkill
- planning-workflowSkill
- pr-review-fanoutSkill
- pull-taskSkill
- reviewSkill
- root-cause-depthSkill
- scan-documentSkill
- self-improvementSkill
- self-verifySkill
- skillifySkill
- task-workflowSkill
- testing-without-mocksSkill
- vault-cloudSkill
- verify-inherited-premiseSkill
The file
We split this skill into the rules it gives an agent
2 rules. 0 are common, 2 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 7 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 50
When invoked, run operations sequentially:
- Distinctive: line 66
Success: All configuration files created.
“Common” is measured: kept by at least 6 in 10 of the 7 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Ralph Wiggum Project Scaffolding
Scaffolds the files needed to run a Ralph Wiggum autonomous control loop on a project. Supports both single-repo projects and multi-repo workspaces.
Overview
The Ralph Wiggum pattern is a while true bash loop that repeatedly feeds the same prompt to Claude Code. Between iterations, files and git history persist while context resets fresh. The agent sees its own previous work and iteratively improves until backpressure gates (tests, lint, build) all pass.
Core philosophy: Backpressure over direction—engineer environments where wrong outputs get rejected automatically.
Multi-Repo Support
Ralph can operate on workspaces containing multiple git repositories. In multi-repo mode:
- A workspace manifest (
.ralph/workspace.md) lists all repos with paths and gate configs - Each repo has its own
.ralph/gates.mdwith repo-specific gate commands - Tasks in
PLAN.mdare annotated with[repo-name]prefix to target specific repos - Branches, commits, and PRs are managed per-repo
- The loop runs from the workspace root directory
Operations
1. Gather Requirements
When: Starting scaffolding, need project info
Implementation: Load operations/gather-requirements.md
Quick summary: Detects single-repo or multi-repo workspace, collects language and gate commands (per-repo in multi-repo mode), and optional external service dependencies via AskUserQuestion.
2. Scaffold Project
When: Requirements gathered, ready to create files
Implementation: Load operations/scaffold-project.md
Quick summary: Checks for existing files, creates from templates with user-provided values. In multi-repo mode, creates workspace manifest and per-repo gate files. Shows next steps.
Prerequisites
- Current directory is a git repository, or contains subdirectories that are git repos (for multi-repo mode)
WriteandAskUserQuestiontools available
End-to-End Flow
When invoked, run operations sequentially:
Input
- Working directory (scanned for git repos to detect workspace mode)
- User responses to AskUserQuestion prompts
Sequence
1. gather-requirements → workspace_type, gate commands, optional services
2. scaffold-project(gathered_values) → created files listOutput
- Success: All configuration files created. User shown file list and next steps (write specs, run plan, run build).
- Partial: Some files skipped (user chose to preserve existing). User shown which files were created vs skipped.
Quick Reference
Invoke skill: User asks to "scaffold ralph wiggum", "set up autonomous loop", or "prepare project for ralph"
Template location: skills/ralph-wiggum/templates/
Three-phase workflow:
- Spec - Human writes requirements in
specs/*.md - Plan - Agent generates
PLAN.mdfrom specs - Build - Agent implements one task per iteration
Backpressure gates: Tests, lint, typecheck, and build must all pass before a task is complete.
Workspace modes:
- Single-repo:
.ralph/gates.mdat project root (original behavior) - Multi-repo:
.ralph/workspace.mdat workspace root + per-repo.ralph/gates.md
Progressive Disclosure
Load only what you need:
operations/gather-requirements.md- Interactive requirements collection (detects workspace type)operations/scaffold-project.md- File creation from templates (supports both modes)references/environment.md- Pre-flight checks and prerequisitestemplates/- File generation templates (gates, services, specs, prompts, workspace)scripts/- Host-side runtime:run-container.sh,loop.sh,lib.sh,setup-network.sh, devcontainer config
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

