oneshotlm
Compare all models on this prompt →

Maze generator & solver · claude-opus-5

Generate a random maze with the recursive-backtracker algorithm on a canvas, then animate an A-star or BFS solver finding the path from start to finish.

⛨ sandboxed Open in new tab ↗
transcript · 4 messages
user

Generate a random maze with the recursive-backtracker algorithm on a canvas, then animate an A-star or BFS solver finding the path from start to finish.

assistant
thinking
I'm creating a single HTML file that combines maze generation with animated solving algorithms—I'll visualize the maze being built, then show BFS and A* pathfinding in action.
write /tmp/ws/index.html (8817 bytes)
toolResult

Successfully wrote 8817 bytes to /tmp/ws/index.html

assistant
thinking
There's a minor issue where the disabled state persists after an abort, but it gets re-enabled when a new run starts, which works fine. For the carving visualization, unvisited cells don't render but their walls do, which creates a full grid outline since all those walls are closed—visually acceptable.

Created `/tmp/ws/index.html` — a single self-contained page: - **Generation**: recursive backtracker (explicit stack), animated with orange carving head and blue live stack. - **Solving**: pick **A\*** (Manhattan heuristic) or **BFS**; frontier in blue, closed set in green, final path in pink, start/goal dots. - **Controls**: maze size (8–60), animation speed, algorithm select, regenerate; live stats (cells explored, path length). No external resources, all CSS/JS inline.

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

modelclaude-opus-5
vendorAnthropic · Claude
harnesspi-agent
generated2026-08-03

EVALUATION

3/5 partial

Maze generator/solver visualization; static image, no idle animation, and mouse/wheel/drag had no effect; keys showed the solved maze legend.

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.

idle motion0.0%
on click0.0%
on drag0.0%
on wheel0.0%
on enter + space36.7%
on arrow keys22.9%
on w/a/s/d0.0%
frame spread28.1 / 255
console errors0
js errors none

METRICS

runtime53.1s
tokens in9.3k
tokens out5.0k
cached3.5k
cost / run$0.1628