Rhythm Desk - heart rate variability analyser for an ECG or PPG trace

Paste a single-channel physiological waveform - an ECG or a PPG/pulse trace, as a timestamped table, optionally with a second respiration channel - and get its cardiac rhythm read two ways. A real biosignal engine runs free in your browser first: it parses the table, detects individual heartbeats with a peak-detection algorithm, and computes exact arithmetic on the result - mean/min/max heart rate, time-domain HRV (SDNN, RMSSD, pNN50, pNN20, coefficient of variation), respiration rate when a respiration channel is present, and eight deterministic signal-quality flags covering flatlines, clipping, irregular sampling, implausible beat-to-beat intervals and short recordings. The raw waveform never leaves the page: only that computed facts table is ever sent, so there is nothing to clip for the model and nothing large to store. Then two lanes work the same numbers - read the rhythm and say what it means for this trace, or turn that read into a short decision brief for a stated purpose (training recovery, stress research, wellness tracking, or a clinical pre-screen note). Every metric the model cites is checked back against what the browser actually computed and labelled if it is not grounded, and every quality flag the engine raised must be answered exactly once. Not a medical device and not a diagnosis. Derived from and crediting the published agent skill @k-dense-ai/neurokit2.

Every public app is built from a security-scanned skill and must pass a clean scan — skill and frontend — before it can be listed. Have a skill of your own? Turn it into an app — or read the step-by-step walkthrough.