Ragdoll Lab

Cinevva's analyzer read this game's design doc and its source, then wrote the brief below. "Make your own" hands it to the game creator as your opening message, so you start from a real spec rather than a one-line idea.

Create a portrait-friendly 2D browser sandbox game titled “Ragdoll Lab,” with a full-screen HTML canvas and dark UI panels. The game is an interactive physics toy rather than a level-based game: let the player spawn cartoon people and props, drag them around, feed them, pose them, and create slapstick chaos, with a running “Chaos score” counter.

Visual presentation: use a dark illustrated laboratory-like background by default, with a nearly black navy gradient and subtle 40px grid lines, a dark gray floor strip along the bottom, and neon muted red/pink accents in the interface. Also support selectable Lab, Night, Sunset, and Forest backgrounds; Night and Forest may use illustrated scene imagery with trees/stars, while Lab/Sunset should have suitable dark gradients. Add alternate Grass, Checker, and Lab floor/map styles. Draw the game in a responsive canvas scaled to the viewport, with touch/pointer input and mouse input.

HUD and controls: at top-left show a bordered panel titled “RAGDOLL LAB” with the subtitle “people, food, physics.” At top-right show “Chaos score,” a large numeric score initially 0, and the current tool label such as “Tool: person average.” Include a Menu toggle and Pause button. The menu has category tabs People, Props, Tools, and World, with expandable button rows. People buttons: Average, Sky spawn, Heavy, Tall, Small, Robot. Props: Crate, Barrel, Knife, Apple, Burger, Bottle, Guts. Tools: Grab, Hold, Stand, Feed, Voice, Erase, Blade, Bomb. World: Night, Sunset, Forest, Lab, Grass map, Checker map, Lab map. Include a Reset lab button and a bottom help panel explaining that the player can spawn items, use illustrated backgrounds, double-tap people for freeze/size controls, drop people from the sky, grab objects, attach items to hands, stand people upright, feed and voice them.

On reset, create four people near the upper/middle area using a mix of average, heavy, and tall types, and five random starter props chosen from crates, barrels, and apples. People should settle into upright poses on the floor. Each person is a simple Verlet-style ragdoll made of nodes and links: head, torso, hips/legs, arms, and feet, with gravity, damping, constrained links, floor collision, and screen-edge clamping. Render them as rounded full-body cartoon figures viewed from the side: flesh-colored limbs and head, colored shirt/torso, dark hair/marking, tiny facial eye/nose/mouth details, circular joints/bone-like highlights, and a small hunger bar above the torso that changes from green to red as hunger rises. Use different scale/body proportions for average, heavy, tall, and small people; the robot variant should have a cool blue-gray look and a ring detail around the head. Dead people should become muted gray-brown. Hunger rises over time, damages health when full, and feeding restores hunger and some health.

Interaction behavior: clicking/tapping with a person tool spawns an upright person at the chosen location; the Sky spawn tool creates an upright person above the tap point and drops them under gravity. Props spawn at the tap point with initial random velocity and fall/bounce on the floor. Render crates as outlined brown boxes with an X, barrels as blue-gray boxes/cylinders, knives as small silver blades with a dark handle and red grip, apples as red circles with a green leaf, burgers as layered buns/filling, bottles as cyan containers, and Guts as pink/red squiggly gore pieces. Every spawn and action should add a small amount to the score; guts and destructive actions are worth more.

Implement Grab as pointer drag: drag a nearby person by moving all of their ragdoll nodes, or drag a nearby item. On release, restore the person’s upright pose; if an item is released near a person, attach it to that person’s hand. Hold should release a held item when used. Stand toggles a nearby person between sitting/ragdoll and a neatly upright standing pose, with a small sound and spoken line. Feed should consume a nearby apple, burger, or bottle when used near a person, reduce hunger substantially, restore a little health, add score, and emit small green particles. Voice should speak a short browser text-to-speech line; hungry people say they are hungry or ask for food, otherwise they say playful lines such as “Hey!” or “That tickles.” Erase removes a nearby prop, or removes a nearby person while producing blood effects. Double-tap a person to open a centered “Person settings” modal showing the selected type and buttons to Freeze/Unfreeze, Make bigger, Make smaller, and Close; freeze stops that person’s simulation, while size can be adjusted within a moderate range.

Violence/physics tools should remain cartoonish: Blade damages the nearest person, creates a burst of red blood particles/stains and several guts props, and knocks a node away. Knife props should damage a nearby person while moving quickly and generate a smaller blood burst plus guts. Bomb creates a large blood burst/stain at the target and applies a radial impulse to nearby ragdoll nodes, people, and props. Blood should appear as irregular dark red splats that slowly fade, with red/pink particles affected by gravity and bouncing slightly on the floor. Add simple synthesized UI/action sounds and use browser speech synthesis for Voice and Stand lines where available.

The simulation should continuously update unless paused, with a Pause button changing to Resume. Include keyboard shortcuts: number keys 1–5 select person types, 6–0 select crate, barrel, knife, apple, and burger, and R resets the lab. Keep the tone playful cartoon laboratory sandbox: dark navy/charcoal space, restrained crimson highlights, colorful rounded ragdolls and props, visible physical motion, no enemies, no win condition, and no invented levels or objectives beyond experimenting and increasing the chaos score.