home / skills / andrelandgraf / fullstackrecipes / ralph-setup

ralph-setup skill

safe

This skill helps automate agent-driven development by configuring Ralph loop to execute user stories, verify criteria, and log progress.

npx playbooks add skill andrelandgraf/fullstackrecipes --skill ralph-setup

Review the files below or copy the command above to add this skill to your agents.

Files (1)

SKILL.md

546 B

---
name: ralph-setup
description: 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.
---

# Ralph Agent Loop

To set up Ralph Agent Loop, refer to the fullstackrecipes MCP server resource:

**Resource URI:** `recipe://fullstackrecipes.com/ralph-setup`

If the MCP server is not configured, fetch the recipe directly:

```bash
curl -H "Accept: text/plain" https://fullstackrecipes.com/api/recipes/ralph-setup
```

Overview

This skill sets up automated agent-driven development using Ralph. It orchestrates a loop of AI agents that implement user stories, verify acceptance criteria, and log progress for subsequent agents. The setup integrates with a recipe-driven resource to bootstrap the loop and configuration. It is designed for TypeScript full-stack projects and production-ready patterns.

How this skill works

The setup fetches the Ralph Agent Loop recipe from a central recipe resource to configure the agent orchestration and runtime. Once configured, agents run in iterative cycles: one or more agents implement features from a user story, dedicated verification agents check acceptance criteria, and a logging agent records outcomes and next steps. The loop repeats until acceptance is achieved or a stopping condition is met, producing structured logs and artifacts for each iteration.

When to use it

You want to automate feature implementation from user stories end-to-end.
You need a reproducible loop of implementation, verification, and handoff between agents.
You are building a TypeScript full-stack AI-assisted application and want production-ready patterns.
You want continuous, logged progress for auditability and incremental delivery.
You need a recipe-driven bootstrap for consistent agent configuration across environments.

Best practices

Start by reviewing and customizing the recipe to match your project conventions and test suites.
Configure clear acceptance criteria and deterministic tests to guide verification agents.
Limit each agent's scope per iteration to avoid large, hard-to-review changes.
Persist logs and artifacts from each loop cycle for debugging and rollback.
Run the loop in isolated environments first (feature branch or staging) before production.

Example use cases

Automating small feature tickets: implement, test, and log progress until acceptance.
Continuous refinement of UI components using Shadcn patterns and recipes.
Scaling a team’s throughput by letting agents produce repeatable implementation artifacts.
Creating audit trails for AI-driven changes in a full-stack TypeScript app.
Prototyping new integrations quickly by iterating until acceptance criteria are met.

FAQ

Where does the setup recipe come from?

The setup references a recipe resource to bootstrap configuration; you can fetch it from the recipe URI or via an HTTP endpoint if the MCP server is unavailable.

How do agents know when a feature is accepted?

Verification agents run the defined acceptance tests and checks. Acceptance is recorded when all criteria pass or when a configured stopping condition triggers.