cth9191

Cloneify — Data skill for Claude Code

Data community

Claude Code skill: turn 'I want to clone X' into a paste-ready build prompt via a gated scope→research→grilled+Codex-reviewed-plan→handoff pipeline.

How to install Cloneify

This entry records only its repository, not the path inside it, so there is no exact command to give. Open cth9191/cloneify and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Cloneify does

Claude Code skill: turn 'I want to clone X' into a paste-ready build prompt via a gated scope→research→grilled+Codex-reviewed-plan→handoff pipeline.

Alternatives in Data

HTML Video — Programmatic video for coding agents — HTML to video on your laptop 4.5k ★
Token Dashboard — See where Claude Code is burning tokens - turn raw JSONL transcripts into local cost analytics, hotspot views 669 ★
Graph Engineering — Graph engineering for AI agents: the 9-stage knowledge-graph pipeline (translated from SEU's graduate course) 452 ★

README

cloneify

**Turn "I want to clone X" into a single paste-ready prompt that builds it.**

`cloneify` is a [Claude Code](https://claude.com/claude-code) skill that conducts a four-phase, human-gated pipeline for cloning a piece of software. You stay in the driver's seat — every phase stops for your go-ahead — and the pipeline composes two other skills rather than reinventing them.

The real deliverable is **Phase 3's goal-prompt**: one self-contained block you paste into a fresh Claude Code session to actually build the clone. Phases 0–2 exist to make that prompt *correct* — scoped, researched, and adversarially hardened.

The four phases

Phase	Name	What happens
0	Scope	One-question-at-a-time interview: platform, must-have features, hard constraints (the reason to clone), stack, explicit out-of-scope. → `scope.md`
1	Research	Invokes a deep-research skill to understand the target deeply — real feature set, technical approach, existing clones, and the landmines most likely to sink the build. → `research.md`
2	Plan	Invokes grill-me-codex: Claude interviews you to lock every open decision, then OpenAI Codex adversarially reviews the plan in a read-only sandbox until `APPROVED`. → `PLAN.md` + `PLAN-REVIEW-LOG.md`
3	Handoff	Emits `GOAL-PROMPT.md` — a self-contained build prompt (full plan inlined) plus a matching `/goal` completion condition so Claude Code builds autonomously until it's met.

**Gates are hard stops.** Each phase produces an artifact, shows it to you, asks the gate question, and waits. You can always say "go back" and revise the prior phase.

Why it works

Scope before research — research has a target, and the plan has guardrails.
Research before planning — decisions rest on sourced facts, not vibes.
Grill + cross-model review before building — the #1 failure is building the wrong thing (fixed by the gri