Ralph Loop
Run an automated Ralph loop so agents keep implementing user stories until every acceptance criterion passes.
Overview
Ralph Loop is an agent skill most often used in Build (also Validate, Ship) that configures automated agent-driven development loops against user stories until all acceptance criteria pass.
Install
npx skills add https://github.com/andrelandgraf/fullstackrecipes --skill ralph-loopWhat is this skill?
- End-to-end Ralph setup: stories → agent loop → pass/fail on acceptance criteria
- Ordered cookbook chain via Fullstack Recipes (agent-setup, user-stories-setup, using-user-stories, then Ralph loop)
- JSON user stories with testable acceptance criteria agents can verify
- Automated re-run loop until all stories pass
- Assumes prior AI agent configuration (Cursor, Copilot, Claude Code) with project patterns
- 4 prerequisite cookbook recipes in ordered sequence (agent-setup, user-stories-setup, using-user-stories, Ralph agent lo
Adoption & trust: 3.5k installs on skills.sh; 17 GitHub stars; 1/3 security scanners passed (skills.sh audits).
What problem does it solve?
You have features described in chat but no repeatable loop that drives agents to implement and verify acceptance criteria until work is actually done.
Who is it for?
Solo builders who already use structured user stories and want hands-off agent iteration until criteria pass.
Skip if: One-off fixes, exploratory spikes without stories, or teams that have not completed agent-setup and user-stories recipes first.
When should I use this skill?
User wants automated agent-driven development with user stories and a loop until all acceptance criteria pass, after Fullstack Recipes prerequisites.
What do I get? / Deliverables
After the cookbook chain, you get a Ralph agent loop tied to JSON user stories so agents iterate until every acceptance criterion passes—then you extend stories or move to the next feature set.
- Ralph agent loop configuration
- JSON user stories with verifiable acceptance criteria
- Iterative implementation until all stories pass
Recommended Skills
Journey fit
Spans multiple journey phases - primary shelf plus alternate fits below.
Canonical shelf is Build/PM because the loop’s core job is sustained feature delivery against structured stories, even though stories and verification span earlier and later phases. PM subphase fits story-driven iteration, tracking acceptance criteria, and orchestrating agent work as a product delivery rhythm—not a one-off codegen snippet.
Where it fits
Turn a feature idea into JSON user stories with testable acceptance criteria before agents touch the codebase.
Run the Ralph loop so agents implement stories and re-run until every criterion is marked passing.
Wire agent-setup patterns from Fullstack Recipes so the loop uses consistent project guidelines and MCP context.
Use passing acceptance criteria as the exit gate before you call a feature done or merge.
How it compares
Use instead of manual re-prompting after each partial implementation when you already commit to story-based acceptance criteria.
Common Questions / FAQ
Who is ralph-loop for?
Indie and solo builders running Claude Code, Cursor, or similar agents who want Ralph-style automated loops over JSON user stories with testable acceptance criteria.
When should I use ralph-loop?
In Validate when you are formalizing scope as stories; in Build when you want agents to implement and re-run until criteria pass; in Ship when acceptance checks act as lightweight verification gates—not for unstructured chat-only builds.
Is ralph-loop safe to install?
Treat it like any workflow that runs agents against your repo: review the Security Audits panel on this Prism page and inspect Fullstack Recipes sources before enabling automated loops on production code.
SKILL.md
READMESKILL.md - Ralph Loop
# Ralph Loop Complete setup for automated agent-driven development. Define features as user stories with testable acceptance criteria, then run AI agents in a loop until all stories pass. ## Prerequisites Complete these recipes first (in order): ### AI Coding Agent Configuration Configure AI coding agents like Cursor, GitHub Copilot, or Claude Code with project-specific patterns, coding guidelines, and MCP servers for consistent AI-assisted development. ```bash curl -H "Accept: text/markdown" https://fullstackrecipes.com/api/recipes/agent-setup ``` ## Cookbook - Complete These Recipes in Order ### User Stories Setup Create a structured format for documenting feature requirements as user stories. JSON files with testable acceptance criteria that AI agents can verify and track. ```bash curl -H "Accept: text/markdown" https://fullstackrecipes.com/api/recipes/user-stories-setup ``` ### Working with User Stories Document and track feature implementation with user stories. Workflow for authoring stories, building features, and marking acceptance criteria as passing. ```bash curl -H "Accept: text/markdown" https://fullstackrecipes.com/api/recipes/using-user-stories ``` ### Ralph Agent Loop Set up automated agent-driven development with Ralph. Run AI agents in a loop to implement features from user stories, verify acceptance criteria, and log progress for the next agent. ```bash curl -H "Accept: text/markdown" https://fullstackrecipes.com/api/recipes/ralph-setup ```