REA
Reverse engineer anything from your terminal or agent with one CLI and MCP server.
The best skill for 3D printing
如果你是作者,可把这段代码贴到 README。徽章会链接回本页。
[](https://mcp.popnx.com/en/skills/mfranzon-print3d)在同一工作中把 AI 连到服务的 MCP
Reverse engineer anything from your terminal or agent with one CLI and MCP server.
Codebase intelligence for TypeScript and JavaScript, exposed as structured MCP analysis tools.
MCP server for terminal commands, file operations, and process management
同类工作,排序与 Skill 列表相同
| 本项 print3d | rea | andrej-karpathy-skills | graphify | MoneyPrinterTurbo | |
|---|---|---|---|---|---|
| 作者 | mfranzon | morluto | multica-ai | Graphify-Labs | harry0703 |
| Skill 数 | 1 | 1 | 1 | 1 | 1 |
| GitHub 星标 | 556 | 465,460 | 233,910 | 149,364 | 144,149 |
| 最近代码更新 | 2026年11月11日 | 2026年11月17日 | 2026年4月20日 | 2026年11月17日 | 2026年11月17日 |
| 许可证 | MIT | MIT | – | Apache-2.0 | MIT |
mfranzon 提供的 1 个 Skill(指令集),AI 读取后照做 · 不是 MCP server
MCP = 手
把 AI 连到外部服务或数据,例如读取 Notion 或开发票。需要有程序或服务器在运行。
Skill = 手册
教 AI 怎么做某项工作的文本文件,不连接任何服务。
有些会重叠:少数 Skill 带有调用工具的脚本,少数 MCP 只提供知识。请从你想完成的工作出发。
Claude CodeCodexCursorGemini CLI
依据开发者 README 和 repo 中的文件夹
GitHub account mfranzon (person), 8 years old
★ 556 stars · code last changed 2026年11月11日
No, plain text files the AI reads and follows
No scripts, so it does not go online by itself
No hooks or writes to machine settings found
Checked automatically from the repo files at commit 5ce50d5 (read only, nothing was run). Not a safety guarantee: read it and decide for yourself
第一步:你用哪个 AI?
选择你用的应用,本页只显示该应用的安装方法
Uploading to claude.ai would fail: the description is over 200 characters (258). Use "Let the AI install it" instead
复制这句话,粘贴到 Claude Code 的对话中
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
重新打开 Claude Code,输入 /print3d,或直接提出相关任务,Claude 会自动选用
Checked against official docs 2026年11月12日 · Source: code.claude.com
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
ChatGPT 桌面应用内置 Codex,Codex 从 ~/.agents/skills 文件夹读取 Skill
在 ChatGPT 桌面应用中打开 Codex,把这句话粘贴到对话中
The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
把 print3d 文件夹放到 ~/.agents/skills/print3d
~ 是你的主文件夹($HOME)
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
不能使用你自己安装的 Skill,只能使用目录中插件自带的 Skill
想在 ChatGPT 中使用这个 Skill,请按上面的方法使用 ChatGPT 桌面应用
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
复制这句话,粘贴到 Codex 的对话中 · 适用于 ChatGPT 桌面应用中的 Codex、Codex CLI 和 IDE 扩展
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Codex 会自动发现新安装的 Skill;如果没有出现,请重启 Codex · 使用方法:在 ChatGPT 中输入 @ 并选择 Skill · 在 Codex CLI 或 IDE 扩展中运行 /skills,或输入 $ 加 Skill 名称
Checked against official docs 2026年11月15日 · Source: learn.chatgpt.com
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
复制这句话,粘贴到 Cursor 的对话中
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
在 Cursor 的 Agent 聊天中输入 / 并选择 print3d
Checked against official docs 2026年11月12日 · Source: cursor.com
Skill 是 AI 会照做的指令。安装前先阅读文件,只安装可信开发者的 Skill。
Mac:按 Command+空格输入 Terminal · Windows:下方命令在 PowerShell 中无法使用,请改用上方的“让 AI 帮你安装”
这条命令把 repo 下载到临时文件夹,只复制含 SKILL.md 的文件夹
正在检查命令 · 通过后才会显示复制按钮
适用于其他能读取 Skill 的应用。这句话列出了 Claude Code、Codex 和 Cursor 的路径,AI 会按你的应用选择 · Copy this sentence into the app’s chat
For one project, paste it into a chat that is open in that project’s folder · The AI always shows you the path before installing
The AI tells you what it will do first and asks your permission at every step
Checked against official docs 2026年11月12日 · Source: code.claude.com, learn.chatgpt.com, cursor.com
其他支持 Agent Skills 的应用(如 Gemini CLI、GitHub Copilot)使用不同文件夹,请查看该应用的指南,再把含 SKILL.md 的文件夹复制过去。
已于 2026年11月11日 对照 Claude Code、Codex 和 Cursor 官方指南核实 skill 文件夹
以下文字来自开发者的 repo,不是 PopNx MCP 的指令
7,452 个字符 · 在 GitHub 打开完整文件
---
name: print3d
description: Model a part in Blender from an idea, a photo of an object, or the shop listing of something you would otherwise buy, slice it for the user's Bambu Lab X2D or P2S or Elegoo Centauri Carbon or Carbon 2, and iterate until it prints clean and matches the photo.
argument-hint: "[what to make, a photo of the object, or a shop link]"
disable-model-invocation: true
---
# /print3d — idea to a printable part
`print3d` below is the print3d command line. When you have the print3d MCP
tools, use those of the same names instead: they return the plate preview and
overlays as images, and the host can ask the user before each one that writes.
A job's files live in the report's `job_dir`, written below as `jobs/<slug>`.
```
idea / photo / text → model.py (Blender) → mesh → DFM + printability
│ ↑ + likeness to the photo
spec.json └──────── advice: what to fix ───┘
│
slice → .3mf → user prints from Studio
```
Nothing here reaches the printer. The last step is a sentence handing the user a
`.3mf`. Printing is theirs to do in Studio, with the plate in front of them.
## 1. Read the brief
Before writing a model for a new job, call the MCP `doctor` tool or run
`print3d doctor`. Resolve any missing or unavailable Blender or Bambu Studio
dependency first. Studio is ready when its executable is found; a failed version
check is informational and does not block modelling. Windows and macOS read
Studio metadata without launching it; Linux uses a best-effort help probe. Blender must pass its
version probe. A null version means the probe could not identify a version;
it does not mean the application is missing. Run the check again after changing
the application paths or installing a dependency.
Size, text, colour, what the part has to do. Ask at most one question, and
only when a missing number makes the part unprintable.
**From a photo of an object:** scaffold (step 2) for the slug, then work through
`references/from-photo.md` before writing the model. It turns the photo into a
`spec.json` of measured, given and assumed dimensions, and adds the `likeness`
check that compares the model with the photo.
**To replace something sold in a shop** (a listing URL or a product name): also
work through `references/from-listing.md`. The listing's stated dimensions scale
its product photos, and the hand-off weighs the print against the price.
**More than one part, more than one colour, or parts that fit together** (a box
and lid, a case in halves, a name in a second colour, a pin in a hole): read
`references/parts-and-fits.md` before writing the model.
## 2. Scaffold
```bash
print3d make "<prompt>"
```
With no model yet this writes `jobs/<slug>/model.py` and stops, naming the path.
Take the slug from that path.
## 3. Write the model
Replace the scaffold's body with the real part. The contract:
- Plain bpy. **1 Blender unit == 1 mm.**
- `params` is in scope: `prompt`, `slug`, `text`, `intent`, `size_mm`, `color`,
`material`, `dims` (name -> mm from `spec.json`) and `fit` (this printer's
clearance per side for `press`, `snug`, `slide`, `free`). Size every mating
feature from `params["fit"]`, never from a guessed number.
- Several parts: model each where it sits when assembled and name it with
`part(objects, name, color=, group=, rotate=, plate=, settings=)`; declare how
two parts mate with `fit(a, b, kind)`. Without `part()` every mesh is joined
into one part.
- Leave the geometry as mesh objects. The harness joins them, removes doubles,
fills holes, recalculates normals, triangulates, centres the part, drops it on
the bed and exports.
- **Orientation is yours.** Studio's auto-orient is off, so what you model is
what prints.
`examples/nameplate.py` is the worked reference. `references/modelling.md` has
the bpy recipes, the gotchas that cost an iteration, and the move each check
asks for.
## 4. Run the loop until it goes green
```bash
print3d make "<prompt>"
```
Models, DFM-checks, slices with the pinned presets of the chosen printer, and
reports. The run is
**red** when `severity` is `error` — exit code 1, `status: needs_work`.
While red: read `advice`, apply the `fix` each note carries to `model.py`, run
again. Each note is phrased as a change to the model, because that is where the
fix lives. `jobs/<slug>/history.jsonl` accumulates what each iteration changed.
Hand over only once it is green.
## 5. Look at the plate
Read the PNG at `plate_png` (each `plates[].png` for several plates), and each
`likeness[].overlay` when there is one.
The numbers go green on parts that are printable and still wrong — mirrored text, a hole in the wrong face, letters too shallow
to read. Look before you hand it over.
`viewer` is `jobs/<slug>/view.html`: the model in 3D, to rotate, zoom, hide parts
and pull them apart. When you have a preview pane (Hermes desktop's
`desktop_preview`, action `open`, with that path as the url), open it there so
the user can turn the part around beside the chat. Otherwise give the path to
open in a browser. `print3d view <slug>` rebuilds it for an older job.
## 6. Hand off
```bash
print3d handoff <slug>
```
writes `jobs/<slug>/handoff/`: a print sheet the user can send on, renders of
the part assembled (and exploded, for several parts), the `.3mf` and an STL per
part. Look at the renders: a colour that does not show, or a part in the wrong
place, is plain there.
Report time, grams and layers, and for several parts the colours and how each
declared fit measured. From a photo, also list which dimensions were
measured, given, listed or assumed, and the unknowns. From a listing, put the
grams and time beside the listing's price. Then give the path:
```
jobs/<slug>/slice/<slug>.3mf
```
That is the project file. The `.gcode.3mf` beside it is a sliced job and
Studio's Prepare tab rejects it as having no geometry. Tell the user to open the
`.3mf`, check the preview, and print from Studio — and offer another iteration
if the preview is not what they wanted.
## Done when
`print3d make` exits 0, no note is at `error`, you have looked at the plate PNGs,
any likeness overlays and the hand-off renders, and the user has the `.3mf` path
and the print sheet.
## The machine
print3d slices for one printer at a time. Before the first run, check which one
with `printer` (`print3d printer` on the command line), and switch when the user
owns another: `printer(choose="p2s-0.4")`. The choice is saved. Every `make`
report names the printer in `printer`; tell the user which printer the `.3mf`
is set up for.
| | X2D (`x2d-0.4`, the default) | P2S (`p2s-0.4`) |
| --- | --- | --- |
| Nozzles | two, 0.4 mm | one, 0.4 mm |
| Build volume | 256 × 256 × 261 mm | 256 × 256 × 256 mm |
| Colours | two print without a filament change | every colour after the first is an AMS filament change, with a purge and a prime tower |
The Elegoo Centauri Carbon (`elegoo-cc-0.4`) and Centauri Carbon 2
(`elegoo-cc2-0.4`) are one-nozzle printers like the P2S, sliced in ElegooSlicer
with Elegoo PLA; each colour after the first is a filament change.
All: 0.20 mm layers, textured PEI plate, PLA (Bambu PLA Basic on the Bambu Lab
printers). **The pinned
processes have supports off** — that is why an overhang note is a real defect
and not a setting to flip.
提问、分享技巧或报告无法使用。按“有帮助”数量排序。