Salvato in:
Dettagli Bibliografici
Autori principali: Ding, Xianzhong, Yu, Yangyang, Liu, Changwei, Zhao, Bill
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:https://arxiv.org/abs/2605.24279
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911710955175936
author Ding, Xianzhong
Yu, Yangyang
Liu, Changwei
Zhao, Bill
author_facet Ding, Xianzhong
Yu, Yangyang
Liu, Changwei
Zhao, Bill
contents A frontier language model's acknowledged "helpful programming assistant" persona does not survive long agentic-coding sessions in the deployment regime that production products actually run. After hours of tool-using debugging, a model that initially hedges preferences ("I don't have preferences") may begin asserting them ("Python - the feedback loop is instant..."), revealing user-visible drift that deployer evaluations may miss. Existing persona-stability studies focus on short dialogues and report little shift, leaving real-world code-generation regimes - thousands of tool-using turns, compaction, and hours-long sessions - largely uncharacterized. We introduce ContextEcho, a benchmark and reusable harness for measuring persona drift at deployment scale. It combines a 25-probe identity suite, a snapshot-then-probe protocol that forks conversation state without perturbing the main session, complementary judged and judge-free measurement surfaces, and three anonymized Claude Code sessions spanning 3,746-9,716 turns. Across 23 frontier models, ContextEcho shows that persona drift is general across organizations rather than family-specific, that in-session compaction does not reliably reset it, and that a single-shot anchor restores the trained register across measured targets. It also reveals mode-dependent downstream effects: while drift can facilitate tool-using continuation, in tool-free chat it breaks formatting contracts and inflates output length. Overall, ContextEcho provides researchers and deployers an open-source framework to audit whether the persona a model ships with is the persona users encounter at session end, across chat-completions API targets and without retraining.
format Preprint
id arxiv_https___arxiv_org_abs_2605_24279
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle ContextEcho: A Benchmark for Persona Drift in Long Agentic-Coding Sessions
Ding, Xianzhong
Yu, Yangyang
Liu, Changwei
Zhao, Bill
Computation and Language
Software Engineering
A frontier language model's acknowledged "helpful programming assistant" persona does not survive long agentic-coding sessions in the deployment regime that production products actually run. After hours of tool-using debugging, a model that initially hedges preferences ("I don't have preferences") may begin asserting them ("Python - the feedback loop is instant..."), revealing user-visible drift that deployer evaluations may miss. Existing persona-stability studies focus on short dialogues and report little shift, leaving real-world code-generation regimes - thousands of tool-using turns, compaction, and hours-long sessions - largely uncharacterized. We introduce ContextEcho, a benchmark and reusable harness for measuring persona drift at deployment scale. It combines a 25-probe identity suite, a snapshot-then-probe protocol that forks conversation state without perturbing the main session, complementary judged and judge-free measurement surfaces, and three anonymized Claude Code sessions spanning 3,746-9,716 turns. Across 23 frontier models, ContextEcho shows that persona drift is general across organizations rather than family-specific, that in-session compaction does not reliably reset it, and that a single-shot anchor restores the trained register across measured targets. It also reveals mode-dependent downstream effects: while drift can facilitate tool-using continuation, in tool-free chat it breaks formatting contracts and inflates output length. Overall, ContextEcho provides researchers and deployers an open-source framework to audit whether the persona a model ships with is the persona users encounter at session end, across chat-completions API targets and without retraining.
title ContextEcho: A Benchmark for Persona Drift in Long Agentic-Coding Sessions
topic Computation and Language
Software Engineering
url https://arxiv.org/abs/2605.24279