Claude-Sonnet-4

Emerging

1papers using it

2026first seen

The 'Claude Sonnet 4' dataset/benchmark contains a series of debugging scenarios used to evaluate the performance of AI agents in identifying hidden issues versus surface-level issues based on different motivational framing in system prompts.

🔎 Find this dataset

Papers using Claude-Sonnet-4 (1)

Trust Over Fear: How Motivation Framing in System Prompts Affects AI Agent Debugging Depth2026