Vibe Manufacturing Blueprint

一行のアイデアから物理パーツへ。プロンプト、設計、発注、検証。

バイブマニュファクチャリングの設計図——チャットボットにパーツを説明し、製造可能なファイルをエクスポートして、加工業者や自宅の3Dプリンターへ送り、届いたものを検証する。記録に残るJeff Deehanの事例が基盤。設計5分、SendCutSendへの156ドルの発注、10日で納品。

こんな方におすすめ

高速な物理プロトタイプを求めるメイカー、ハードウェア愛好家、創業者

所要時間

最初のシンプルなパーツはひと午後

レベル

中級

3ステップでインストール

登録不要。CLIツールも不要です。お使いのAIを選んで、1行コピーして貼るだけです。

Pro — 全531件のスキル

ご自分で入れたい場合は、スキルを購入（$4.99）し、購入完了画面からZIPをダウンロードして、上のディレクトリに展開し、AIのセッションを再起動してください。結果は同じです。すでに購入済みの場合は、AIアシスタントにダウンロードしてもらうこともできます。

パックの内容

SKILL.md（手順書）1点とスターターテンプレート1点が入っています。プレビューはこちらです：

Vibe Manufacturing Blueprint

The idea in one paragraph

Vibe coding asked AI to write software from a feeling; vibe manufacturing asks it to design hardware. You describe a part to a chatbot, it produces a design you can export as a manufacturable file, and you send that file to a fabrication service or your home 3D printer. Ten days later — in the documented case — a steel part arrives in the mail. This skill is the blueprint for that pipeline: prompt, export, order, and verify, with a budget for the trial-and-error this process still demands.

What it is built on

The "vibe manufacturing" pattern was reported on November 1, 2026 (Business Insider via digitaltoday.co.kr): a chatbot designs hardware, and the file goes to a manufacturer or a home 3D printer.
The anchor case: Jeff Deehan, a Michigan real estate developer in East Lansing, asked a chatbot to design a steel part for the GPT-6 Astra voice mode — about five minutes of design, a DXF download, a $156 order to SendCutSend, delivery ten days later.
Amy Zhou and Jake Fitzgerald's open-source text-to-CAD tool has 23,000+ GitHub stars, per the same reporting — the tooling layer is open and active.
The reported limits: complexity and high trial-and-error cost. Simple parts work; complex assemblies still need a human engineer.

Workflow

Write the one-line need (input: problem -> output: requirement sentence). State what the part must do and its key constraints (size, material, load) in one sentence. If you cannot say it in one sentence, the design prompt is not ready.
Design by conversation (input: requirement -> output: design draft). Ask the chatbot for the part, then iterate on dimensions and fit in plain language. Aim for the documented pace — minutes, not days — for a simple part.
Export the manufacturable file (input: design -> output: file, e.g. DXF). Get the design out as a file a fabricator can read. Confirm units, scale, and thickness before you send anything anywhere.
Order or print (input: file -> output: order / print job). Send the file to a fabrication service (as Deehan did with SendCutSend) or slice it for a home 3D printer. Record the quoted cost and lead time next to your budget.
Budget the trial and error (input: quote -> output: budget line). Set aside a multiple of the first order's cost for remakes — the reporting is explicit that trial-and-error runs expensive. One successful part is a data point, not a process.
Verify on arrival (input: delivered part -> output: pass/fail). Measure the part against your requirement sentence: dimensions, fit, material. Log what the chatbot got right and what needed a human fix, and feed that back into step 2 next time.