Boids flocking · qwen3-vl-30b-a3b-instruct
Create a full-screen flocking simulation of boids using separation, alignment, and cohesion, drawn as oriented triangles that steer smoothly and flee from the mouse cursor.
transcript · 4 messages
Create a full-screen flocking simulation of boids using separation, alignment, and cohesion, drawn as oriented triangles that steer smoothly and flee from the mouse cursor.
I'll create a full-screen flocking simulation of boids with separation, alignment, cohesion, and mouse avoidance using HTML5 Canvas. The boids will be drawn as oriented triangles that steer smoothly.
index.html (8484 bytes)Successfully wrote 8484 bytes to index.html
I've created a full-screen flocking simulation of boids with separation, alignment, cohesion, and mouse avoidance. The boids are drawn as oriented triangles that steer smoothly and flee from the mouse cursor. The simulation includes: 1. **Separation**: Boids avoid colliding with nearby boids by calculating steering forces to maintain a minimum distance 2. **Alignment**: Boids match their velocity with neighboring boids to move in the same direction 3. **Cohesion**: Boids move toward the average position of nearby boids to stay together 4. **Mouse avoidance**: Boids flee from the mouse cursor when it comes within a certain radius The boids are represented as green triangles that rotate to face their direction of movement. The entire simulation runs at full screen with smooth animation. To run this, simply open the index.html file in any modern web browser. The boids will automatically start moving, and you can move your mouse around to see them flee from the cursor.
TAKE IT TO CODE
A single self-contained HTML file — copy it and open in any browser, or reuse the prompt with your own pi-agent run.
RUN
EVALUATION
Green triangular boids continuously animate and drift across the screen, but no clear flocking or mouse-flee behavior visible.
Vision grade · anthropic/claude-sonnet-5
MEASUREMENTS
Raw readings from the capture, not a verdict — the grade above is the only judgement. Percentages are how much of the screen changed; the measure under-reports sparse motion on a dark background.