An AI agent skill that turns any subject into a 3D film or app
Visualize any subject in 3D: add the showcase mode, group looks into four families with a default per mode, and show a menu from an empty /viz3d; release 0.7.0
Nguyen Xuan Anbc2fde596b62
diff --git a/.claude-plugin/marketplace.json b/.claude-plugin/marketplace.jsonindex df52fcc..89596bd 100644--- a/.claude-plugin/marketplace.json+++ b/.claude-plugin/marketplace.json@@ -1,6 +1,6 @@{"name": "codepawl",- "description": "viz3d: turn hard-to-picture places, systems and events into a 3D world you can watch or click through",+ "description": "viz3d: visualize anything in 3D, as a film, an app you click through, a lesson, a book or a game","owner": {"name": "CodePawl","url": "https://codepawl.com"@@ -9,7 +9,7 @@{"name": "visualize-in-3d","source": "./plugins/visualize-in-3d",- "description": "Turn hard-to-picture places, systems and events into a 3D world you can watch or click through"+ "description": "Visualize anything in 3D: a place, a system, an object, a creature, a story or an idea, as a film, an app, a lesson, a book or a game"},{"name": "build-minecraft-model-from-image",diff --git a/harness/gemini/viz3d.toml b/harness/gemini/viz3d.tomlindex 6f28652..abdd9c5 100644--- a/harness/gemini/viz3d.toml+++ b/harness/gemini/viz3d.toml@@ -1,8 +1,12 @@-description = "Turn a place, system, object or event into a 3D world with the visualize-in-3d skill"+description = "Visualize anything in 3D with the visualize-in-3d skill"prompt = """Activate the visualize-in-3d skill and follow it from the start for this request:{{args}}-If the first word is a mode, use that shape and drop the word from the subject. The modes are app (live app with a dashboard and a short film), lesson (lesson app), prototype (quick prototype from primitives), embed (live embed on real data), film or doc (documentary film), action (action short), book (illustrated cutaway plate, or a book of several with quizzes), game (playable browser game). Otherwise pick the shape from the subject with the skill's table. If the request is empty, ask what the user wants people to see.+If the request is empty or is "help", list the modes and looks below and ask what the user wants to make 3D. Otherwise, if the first word is a mode, use that shape and drop the word from the subject; if the request names a look, use that look, else the default look for the shape.++Modes: film or doc (documentary film), action (action short), showcase (film or stills that show off one object or model), app (live 3D app with a dashboard and a short film), lesson (a real object built and explained step by step), book (illustrated cutaway plate, or a book with quizzes), game (playable browser game), prototype (quick prototype from primitives), embed (live 3D view inside a dashboard, on real data). With no mode, pick the shape from the subject with the skill's table.++Looks: clean product (ops, toy, lived-in), real and cinematic (terrain, documentary, white model, layer build), technical (point cloud, plate), stylized art (pixel, ASCII, glow)."""diff --git a/harness/opencode/viz3d.md b/harness/opencode/viz3d.mdindex 888e40d..ab6112d 100644--- a/harness/opencode/viz3d.md+++ b/harness/opencode/viz3d.md@@ -1,9 +1,13 @@----description: Turn a place, system, object or event into a 3D world with the visualize-in-3d skill+description: Visualize anything in 3D with the visualize-in-3d skill---Load the visualize-in-3d skill and follow it from the start for this request:$ARGUMENTS-If the first word is a mode, use that shape and drop the word from the subject. The modes are app (live app with a dashboard and a short film), lesson (lesson app), prototype (quick prototype from primitives), embed (live embed on real data), film or doc (documentary film), action (action short), book (illustrated cutaway plate, or a book of several with quizzes), game (playable browser game). Otherwise pick the shape from the subject with the skill's table. If the request is empty, ask what the user wants people to see.+If the request is empty or is "help", list the modes and looks below and ask what the user wants to make 3D. Otherwise, if the first word is a mode, use that shape and drop the word from the subject; if the request names a look, use that look, else the default look for the shape.++Modes: film or doc (documentary film), action (action short), showcase (film or stills that show off one object or model), app (live 3D app with a dashboard and a short film), lesson (a real object built and explained step by step), book (illustrated cutaway plate, or a book with quizzes), game (playable browser game), prototype (quick prototype from primitives), embed (live 3D view inside a dashboard, on real data). With no mode, pick the shape from the subject with the skill's table.++Looks: clean product (ops, toy, lived-in), real and cinematic (terrain, documentary, white model, layer build), technical (point cloud, plate), stylized art (pixel, ASCII, glow).diff --git a/plugins/visualize-in-3d/.claude-plugin/plugin.json b/plugins/visualize-in-3d/.claude-plugin/plugin.jsonindex 64cf7db..08cbd50 100644--- a/plugins/visualize-in-3d/.claude-plugin/plugin.json+++ b/plugins/visualize-in-3d/.claude-plugin/plugin.json@@ -1,7 +1,7 @@{"name": "visualize-in-3d",- "version": "0.6.1",- "description": "Turn hard-to-picture places, systems, objects and events into a 3D world you can watch, learn from or click through: a live three.js app with a dashboard, a lesson app, a documentary film, or an illustrated cutaway book, built from real layouts and models made from real references.",+ "version": "0.7.0",+ "description": "Visualize anything in 3D: places, systems, objects, products, creatures, stories and events, as a documentary or action film, a showcase, a live three.js app with a dashboard, a lesson, an illustrated cutaway book or a playable game, in twelve looks across four families, built from real layouts and models made from real references.","author": {"name": "CodePawl","url": "https://codepawl.com"diff --git a/plugins/visualize-in-3d/skills/visualize-in-3d/SKILL.md b/plugins/visualize-in-3d/skills/visualize-in-3d/SKILL.mdindex b2ca736..3229154 100644--- a/plugins/visualize-in-3d/skills/visualize-in-3d/SKILL.md+++ b/plugins/visualize-in-3d/skills/visualize-in-3d/SKILL.md@@ -1,18 +1,18 @@---name: visualize-in-3d-description: Turn anything hard to picture into a 3D world people can watch or play with - an operation or system (airport, data center, port, factory, hospital), an event spreading over real land (wildfire, flood), a story in time (a historical event, a case), characters or machines in action (a fight, a chase, a launch), or how a thing works. Builds a live three.js app with a dashboard and a 30 s film, a lesson app, a quick no-Blender prototype, a live embed that runs on real data inside an existing dashboard, a documentary film, an action short, an interactive cutaway plate or book, or a playable web game. Looks include clean ops world, toy and lived-in isometric, point cloud, photoreal terrain, dark and white model documentary, pixel diorama, cutaway plate, ASCII showcase and layer build film. Models come from ready sources or Blender builds made from real references, and the skill checks its own frames before showing anyone. Use for "visualize how X works", "show this event on a 3D map", "a live 3D dashboard of X", "a 3D lesson on how X works", "a documentary-style 3D explainer", "a quick 3D prototype", "two creatures fight in 3D", "app nhìn như game", "làm phim 3D về", "cho 2 con đánh nhau", "pixel battle", "theme point cloud", "kiểu wildfire tool".+description: Visualize anything in 3D - a place, an operation or system (airport, data center, port, factory, hospital), an object or product, a creature or character, a story or event in time, something spreading over real land, or how a thing works. Builds a live three.js app with a dashboard and a 30 s film, a lesson app, a quick no-Blender prototype, a live embed on real data, a documentary film, an action short, a showcase film or stills of one object, an interactive cutaway plate or book, or a playable web game. Looks come in four families (clean product, real and cinematic, technical, stylized art), from clean ops world and isometric dioramas to photoreal terrain, white model, point cloud, cutaway plate, pixel, ASCII, glow and layer build. Models come from ready sources or Blender builds made from real references, and the skill checks its own frames before showing anyone. Use for "visualize X in 3D", "make a 3D film of X", "show this event on a 3D map", "a live 3D dashboard of X", "a 3D lesson on how X works", "show off this model", "a quick 3D prototype", "two creatures fight in 3D", "dựng 3D cái này", "app nhìn như game", "làm phim 3D về", "cho 2 con đánh nhau", "pixel battle", "theme point cloud", "kiểu wildfire tool".license: MIT----# Visualize in 3D (Dựng thế giới 3D để thấy mọi thứ vận hành)+# Visualize in 3D (Dựng mọi thứ thành 3D)If `references/local.md` exists, read it first: it sets the folders and house rules for this machine.-Take any subject and make it visible in 3D: models from real references, a seeded simulation or a scripted timeline+Take any subject and make it 3D: models from real references, a seeded simulation or a scripted timelinedriving them, a look chosen for the subject, and a film checked frame by frame. The user can also start it with-`/viz3d <mode> <subject>` (modes: app, lesson, prototype, embed, film, action, book, game).+`/viz3d <mode> <subject>` (modes: app, lesson, prototype, embed, film, action, showcase, book, game).Paths use `{{WORK_DIR}}` for the folder where projects live (the current directory unless the user names one) and-`{{SKILL_DIR}}` for this skill's folder. Write paths with forward slashes on every platform.+`{{SKILL_DIR}}` for this skill's folder, with forward slashes on every platform.## Before you start@@ -39,15 +39,15 @@ Read what the user wants people to see, then pick one shape (or combine two) and| Something spreading over real land | Terrain app with a time scrubber | `references/looks.md` (photoreal terrain), `examples/wildfire-valley.md` || A story, case or piece of history | Documentary film | `references/documentary.md`, `references/film.md`, `references/67js.md` || Characters, creatures or machines doing something | Action short | `references/action.md`, `references/67js.md` |+| One object, product, model or creature to show off | Showcase film or stills | `references/looks.md` (Dither + ASCII showcase, Gradient glow ASCII), `references/film.md`, `references/67js.md` || One object cut open or exploded | Cutaway film, or an app with a section view | `references/film.md`, `references/looks.md`, `references/67js.md` || One big object or building with many rooms, as a page or a book | Book: interactive illustrated cutaway plate, with quizzes | `references/book.md`, `references/looks.md` || A game people play in the browser (steer a creature, survive, grow) | Playable web game, bot-balanced | `references/game.md` || A demo for leaders in about an hour | Quick prototype from three.js primitives | `references/live-embed.md` || A live view inside an existing dashboard or product | Embed (web component or iframe) on real data | `references/live-embed.md`, `scripts/mock-feed.mjs` |-Pick the look in `references/looks.md` (Choosing); clean ops world is the default for a system that runs. Study a-look the user shows from a reference video with timestamped frame sheets-(`scripts/contact-sheet.sh <video> <folder> <fps>`), then write it up as a new section in `looks.md`.+Pick the look in `references/looks.md` (Choosing: four families, a default per shape). Study a look the user shows from+a reference video with frame sheets (`scripts/contact-sheet.sh <video> <folder> <fps>`) and add it to `looks.md`.## 2. Write the plan filediff --git a/plugins/visualize-in-3d/skills/visualize-in-3d/references/looks.md b/plugins/visualize-in-3d/skills/visualize-in-3d/references/looks.mdindex 3d3c145..97559f2 100644--- a/plugins/visualize-in-3d/skills/visualize-in-3d/references/looks.md+++ b/plugins/visualize-in-3d/skills/visualize-in-3d/references/looks.md@@ -1,10 +1,10 @@# LooksEach topic file names one look. The look sets the world's rendering, the UI skin and what the film shows; the data-model, the dashboard parts and the film's click-driven shape stay the same. Reading map: Toy isometric (default),+model, the dashboard parts and the film's click-driven shape stay the same. Reading map: start with Choosing at the end (four families and the default look per shape), then Toy isometric,Point-cloud survey, Photoreal terrain analysis, Dark documentary, Pixel diorama, Illustrated cutaway plate,Lived-in isometric, Dither + ASCII showcase, Gradient glow ASCII, Clean ops world, White model documentary, Layer-build film, Choosing.+build film.Reference stills or clips for a look are optional. If you have some, save them under `{{WORK_DIR}}/references/<slug>/`and point the briefs at them; they are for study only and never ship in the app or the film. Without them, the@@ -318,6 +318,28 @@ one is a finished, cinematic frame. No labels and no narration: the picture alon## Choosing+The looks fall into four families. Pick the family from what the piece is for, then the look from the subject.++| Family | Looks | Pick it when |+|---|---|---|+| Clean product | Clean ops world, Toy isometric, Lived-in isometric | The piece should read like a real product or a friendly diorama: apps, lessons, prototypes, embeds |+| Real and cinematic | Photoreal terrain, Dark documentary, White model documentary, Layer build film | It should feel real and filmed: land, events, cases, a place being built |+| Technical | Point-cloud survey, Illustrated cutaway plate | It should read like a survey or an engineering plate: terrain from above, one object cut open |+| Stylized art | Pixel diorama, Dither + ASCII showcase, Gradient glow ASCII | The look is part of the art: battles, games, showing off a model, a creature as art |++Default look per shape, when the user names none:++| Shape | Default look |+|---|---|+| Live app, quick prototype, live embed | Clean ops world |+| Lesson app | Clean ops world, close to the object |+| Terrain app | Photoreal terrain |+| Documentary film | Dark documentary; white model documentary when places and action must be seen clearly |+| Action short | Pixel diorama for creature battles; otherwise the look of the world the action happens in |+| Showcase film or stills | Dither + ASCII showcase with the model rendered clean |+| Book | Illustrated cutaway plate |+| Game | Pixel diorama or toy isometric |+A build-up or breakdown of a place, where the wow is watching it become real, goes layer build film. A case that needs its action and places clearly seen can go white model documentary; dark documentary suits a case that lives onwhat stays hidden. Clean ops world is the default for a system that runs (an operation with units, tasks and flows). Toy isometric (amore playful diorama on a slab, stronger colour) stays as an alternative when the user asks for it or the subject isdiff --git a/plugins/visualize-in-3d/skills/viz3d/SKILL.md b/plugins/visualize-in-3d/skills/viz3d/SKILL.mdindex e739644..42c8441 100644--- a/plugins/visualize-in-3d/skills/viz3d/SKILL.md+++ b/plugins/visualize-in-3d/skills/viz3d/SKILL.md@@ -1,7 +1,7 @@---name: viz3d-description: Start the visualize-in-3d skill from a command, with an optional mode first (app, lesson, prototype, embed, film, action, book, game) and then the subject, for example "/viz3d film the 1906 San Francisco earthquake".-argument-hint: "[app|lesson|prototype|embed|film|action|book|game] <subject>"+description: Start the visualize-in-3d skill from a command, with an optional mode and look first and then the subject, for example "/viz3d film white model the 1906 San Francisco earthquake". With no request it shows the menu of modes and looks.+argument-hint: "[app|lesson|prototype|embed|film|action|showcase|book|game] [look] <subject>"disable-model-invocation: true---@@ -13,22 +13,31 @@ $ARGUMENTSIn tools that do not fill in the line above, the request is the rest of the user's message.-1. Load the `visualize-in-3d` skill and follow it from the start. It is installed next to this one+1. If the request is empty or is `help`, show the menu below as it is, then ask one question: what do you want to+ make 3D? Stop there.+2. Otherwise load the `visualize-in-3d` skill and follow it from the start. It is installed next to this one(`../visualize-in-3d/SKILL.md`).-2. If the first word of the request is a mode, use that shape and drop the word from the subject:-- | Word | Shape |- |---|---|- | `app` | Live app with a dashboard and a short film |- | `lesson` | Lesson app that builds a real object step by step |- | `prototype` | Quick prototype from primitives, no Blender |- | `embed` | Live embed for an existing dashboard, on real data |- | `film` or `doc` | Documentary film |- | `action` | Action short |- | `book` | Illustrated cutaway plate of one subject, or a book of several with quizzes |- | `game` | Playable browser game, balanced by bot playtests |-- Otherwise pick the shape from the subject with the skill's table.-3. If the request names a look (ops, toy, lived-in, point cloud, terrain, documentary, white model, pixel, plate, ASCII,- glow, layer build), use the matching section of `references/looks.md`. Otherwise the skill picks one.-4. If the request is empty, ask one question: what do you want people to see?+3. If the first word is a mode, use that shape and drop the word from the subject. Otherwise pick the shape from the+ subject with the skill's table.+4. If the request names a look, use the matching section of `references/looks.md`. Otherwise use the default look for+ the shape (looks.md, Choosing).++## Menu++**Modes**, by what you want people to do with it:++| Watch it | Explore it | Put it in your product |+|---|---|---|+| `film` or `doc`: documentary film of a story or event | `app`: live 3D app with a dashboard and a 30 s film | `prototype`: quick prototype from primitives, no Blender |+| `action`: action short, things fighting, chasing, launching | `lesson`: a real object built and explained step by step | `embed`: live 3D view inside your dashboard, on your data |+| `showcase`: a film or stills that show off one object or model | `book`: illustrated cutaway plate, or a book with quizzes | |+| | `game`: playable browser game | |++**Looks**, in four families:++| Clean product | Real and cinematic | Technical | Stylized art |+|---|---|---|---|+| ops, toy, lived-in | terrain, documentary, white model, layer build | point cloud, plate | pixel, ASCII, glow |++Examples: `/viz3d app a container port`, `/viz3d film white model the Chernobyl accident`,+`/viz3d showcase glow a cockroach`, `/viz3d game pixel a worm that eats a city`.