Source record

Published by arXiv cs.HC (T1 source). The original is at https://arxiv.org/abs/2607.20773.

Pipeline notes

The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.

HARP is a research platform that places participants in controlled scenarios with live, configurable LLM agents. It logs prompt composition behavior (keystroke timing, deletions, pauses), triggers surveys at predefined moments, and allows researchers to control prompts, model parameters, and experimental conditions. The authors illustrate it with a study on how technical specificity and response length affect retention.

A purpose-built testbed for measuring not just what users ask AI but how they compose and revise prompts — a behavioral layer most usability studies miss.

Cited by

No citations on record.