Skip to content

Author

Zhengyuan Liu

1 paper indexed here

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Preprint Aug 2026

ScrambleToolBench: Agents Search Exhaustively Even When Their Own Map Points to the Next Step

ScrambleToolBench is introduced, an interactive terminal benchmark designed to isolate behavioral reasoning by removing semantic cues and enforcing a continuous task curriculum, which requires agents to uncover hidden tool behaviors entirely through trial-and-error interaction.

Vernon Toh, Navonil Majumder, Zhengyuan Liu et al. · 1 citation