The user wants a Gray-Scott reaction-diffusion simulation that:
1. Runs on a canvas
2. Grows organic Turing patterns in real time
3. Allows mouse seeding for new growth
4. All in a single index.html file with inline CSS and JS
5. No external resources, CDNs, or network requests

The Gray-Scott model is a reaction-diffusion system that produces Turing patterns. It uses two chemicals, U and V, that diffuse and react with each other. The equations are:

∂U/∂t = Du * ∇²U - UV² + f * (1 - U)
∂V/∂t = Dv * ∇²V + UV² - (f + k) * V

Where:
- U and V are concentrations
- Du and Dv are diffusion coefficients
- f is the feed rate
- k is the kill rate

I'll create a complete implementation with:
- Canvas for rendering
- Simulation loop
- Mouse interaction for seeding
- Color mapping for visualization
- Controls for parameters (optional but helpful)

Let me write this as a single HTML file.

I've created a complete Gray-Scott reaction-diffusion simulation. Let me verify the file was created correctly.

I've created a complete, self-contained Gray-Scott reaction-diffusion simulation in `/tmp/ws/index.html`. The simulation features:

**Core Simulation:**
- Gray-Scott reaction-diffusion equations with two chemicals A and B
- Laplacian diffusion with periodic boundary conditions (wrapping)
- Real-time organic Turing pattern formation

**Interactive Features:**
- Click and drag on canvas to seed new growth (adds chemical B)
- Touch support for mobile devices
- Smooth line interpolation when dragging

**Controls:**
- Feed rate and kill rate sliders (the key parameters for different patterns)
- Diffusion coefficients for both chemicals
- Adjustable brush size for seeding
- Simulation speed control
- 6 preset patterns (Coral, Mitosis, Spots, Stripes, Worms, Fingerprints)
- 4 color modes (Rainbow, Fire, Ocean, Matrix)
- Reset and Clear buttons

**Technical Details:**
- 250x250 simulation grid scaled up to 500x500 canvas
- Multiple simulation steps per frame for faster evolution
- FPS counter for performance monitoring
- All CSS and JS inline, no external dependencies

The simulation should produce beautiful organic Turing patterns that evolve in real-time, and users can interact by clicking/dragging to create new growth points.