ClawTrojan Benchmark

What is ClawTrojan?

A reusable benchmark for measuring multi-step trojan attacks that establish persistent control through local agent workspace state.

Released

2026-05-29

Evaluates

General AI, Language & Knowledge, cs.CR

Openness

unknown

Importer

Claire Radar

Review status

ai-reviewed

Reported scores

0

Source provenance

• Original evidence https://arxiv.org/abs/2605.31042

ClawTrojan paper and code

• Paper https://arxiv.org/abs/2605.31042
• Code https://github.com/RUC-NLPIR/ClawTrojan

Which sources cite ClawTrojan?

• ClawTrojan

Related General AI, Language & Knowledge, cs.CR benchmarks

• $\tau^\tau$-Bench: An Environment for End-To-End, Realistic Agent Construction
• 3DCodeBench
• 5-Dialects-BN
• A Large-Scale Dataset and Benchmark
• A3S-Bench
• AARRI-Bench
• ABC-Bench
• ActBench