HarnessRisk Benchmark

What is HarnessRisk?

HarnessRisk evaluates agent harness safety across six lifecycle phases with 128 sandboxed cases. It measures Utility, Attack Success Rate, Persistence, and…

Released

2026-08-18

Evaluates

Agent safety, Agents, Cybersecurity, Persistent-state safety, Prompt-injection resistance, Safety, Safety & Trustworthiness, Tool-use safety

Openness

unknown

Importer

Claire Radar

Review status

source-reviewed

Reported scores

0

Source provenance

• Original evidence https://arxiv.org/abs/2608.17597

HarnessRisk paper, code and dataset

• Paper https://arxiv.org/abs/2608.17597
• Code https://github.com/Baiyajing/HarnessRisk
• Dataset https://huggingface.co/datasets/YajingB/HarnessRisk
• Website https://baiyajing.github.io/harness-risk/

Which sources cite HarnessRisk?

• HarnessRisk

Related Agent safety, Agents, Cybersecurity, Persistent-state safety, Prompt-injection resistance, Safety, Safety & Trustworthiness, Tool-use safety benchmarks

• $\tau^\tau$-Bench: An Environment for End-To-End, Realistic Agent Construction
• A3S-Bench
• ActBench
• Adaptive Adversaries
• AdversaBench
• AeroCopilotBench
• AFT-Bench
• Agent App Benchmark