oneshotlm
Compare all models on this prompt →

Arpeggiator pad · gemini-3.1-flash-lite

Using the pre-provided tone.min.js (global Tone), build a clickable pentatonic pad grid that triggers a synth voice with reverb, plus a play/stop button that runs an arpeggiator sequence at an adjustable tempo. All sound synthesized, no audio files.

⛨ sandboxed Open in new tab ↗
transcript · 6 messages
user

Using the pre-provided tone.min.js (global Tone), build a clickable pentatonic pad grid that triggers a synth voice with reverb, plus a play/stop button that runs an arpeggiator sequence at an adjustable tempo. All sound synthesized, no audio files.

assistant
thinking
**Defining the Core Structure**

I'm now focused on the foundational structure, planning to use `index.html` for a single-page web app. My aim is to include `tone.min.js` which will give me a global `Tone` object. The essential features will center on a pentatonic pad grid and an arpeggiator control with play/stop.


bash {"command":"ls -F"}
toolResult

tone.min.js

assistant
write index.html (3215 bytes)
toolResult

Successfully wrote 3215 bytes to index.html

assistant

The web application is ready in `index.html`. It includes: - A logic-driven pentatonic 5-note scale ($C4, D4, E4, G4, A4$). - A 20-pad grid that triggers the synth voice. - A global `Tone.Reverb` effect processed through the `Tone.PolySynth`. - An adjustable tempo arpeggiator controlled by `Tone.Transport`. - Uses only the local `tone.min.js` file. You can now open `index.html` in an appropriate browser environment to use the instrument.

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

modelgemini-3.1-flash-lite
vendorGoogle
harnesspi-agent
generated2026-07-22

EVALUATION

2/5 broken

Static pentatonic pad grid with play button; no visual feedback to click, drag, wheel, or keyboard input.

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 + space0.0%
on arrow keys0.0%
frame spread21.6 / 255
console errors0
js errors none

METRICS

runtime7.6s
tokens in6.5k
tokens out1.6k
cached0
cost / run$0.0040