AI BENCHMARK PROFILE

ContextEcho

General AIKnowledge & Reasoning

ContextEcho measures persona drift in long agentic-coding sessions, combining a 25-probe identity suite, snapshot-then-probe protocol, and three anonymized Claude Code sessions.

Released: 2026-05-22
Readiness: Paper only
Primary field: General AI

Why it matters

Persona drift can affect user trust and model reliability in real-world deployment, but existing evaluations may miss it. ContextEcho provides a framework for auditing persona consistency across long sessions.

Motivation

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.

Primary resources

Paper ↗Hugging Face ↗

Benchmark Radar records only publicly supported details and links back to primary sources for verification.