gepetto

Generates sectionized implementation plans from markdown specs through research, interviews, and multi-LLM review.

2|Updated Jan 10, 2026
One-click install
npx skills add https://github.com/Shubh2310-developer/ENGUNITYCORE --skill gepetto-shubh2310-developer
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: gepetto
Source: https://github.com/Shubh2310-developer/ENGUNITYCORE/tree/main/.claude/skills/gepetto
Command: npx skills add https://github.com/Shubh2310-developer/ENGUNITYCORE --skill gepetto-shubh2310-developer

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Vague feature ideas often lead to premature coding, missed edge cases, and scope creep. This Skill transforms rough markdown specs into detailed, reviewed implementation plans before any code is written. ## Core Features & Use Cases - Structured Planning Pipeline: Runs a 17-step workflow covering research, stakeholder interview, spec synthesis, plan generation, and section splitting. - External Multi-LLM Review: Sends the plan to Gemini CLI and Codex CLI for independent architectural feedback, then integrates accepted suggestions. - Parallel-Ready Sections: Splits the plan into self-contained section files with dependencies and acceptance criteria, plus ready-to-run prompts for ralph-loop and Ralphy autonomous execution. - Use Case: You have a rough idea for an OAuth2 authentication system. Run the skill on your spec file, answer interview questions, and receive a reviewed plan split into implementation sections you can execute manually or autonomously. ## Quick Start Create a markdown spec file describing your feature in a planning directory, then invoke the skill with that file path, for example /gepetto @planning/auth-spec.md, and answer the interview questions.

Frequently Asked Questions about gepetto

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

FAQPage Schema
How do I create an implementation plan from a rough feature idea?

Write your idea as a markdown spec file, even just bullet points, then invoke the skill with that file path. It runs research, an interview, spec synthesis, and plan generation, producing a complete implementation plan split into self-contained sections.

How does multi-LLM plan review work with Gemini and Codex?

The skill sends the generated plan to Gemini CLI and Codex CLI in parallel, asking each to identify footguns, security issues, and missing considerations. You then decide which suggestions to integrate, with decisions documented in claude-integration-notes.md.

Can I resume an interrupted planning session?

Yes, re-run the skill with the same spec file path. It detects existing planning files like claude-research.md or claude-plan.md and resumes from the appropriate step instead of starting over.

What happens if Gemini CLI or Codex CLI is not installed?

The external review step is skipped or partially completed. If one CLI is available, only that review is written; if neither is installed, the workflow continues without external review.

When should I not use a structured planning workflow?

Skip it for simple bug fixes, one-file changes, or when requirements are already crystal clear. The research, interview, and review phases add overhead that only pays off for complex or ambiguous features.