TraceDance: An Automated System for Building Agent Behavior Benchmarks from Real-World Agent Deployment Traces

arXiv:2609.33295 · cs.AI, cs.CL · Submitted 2026-09-27 · Read on arXiv

cs.AI, cs.CL

Submitted: 2026-09-27

Updated: 2026-09-27

Comments: 34 pages, 7 figures. Project website: https://zhishanq.github.io/TraceDance/

Code: https://github.com/ZhishanQ/TraceDance

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Benchmark — 131× in this paper · explained in 2 episodes
• Trace — 61× in this paper · explained in 2 episodes
• Verification — 8× in this paper · explained in 2 episodes
• Grounding — 7× in this paper · explained in 3 episodes
• Benchmarking — 6× in this paper · explained in 3 episodes
• Adaptation — 5× in this paper · explained in 1 episode
• Clustering — 3× in this paper · explained in 2 episodes
• LLM judge — 3× in this paper · explained in 2 episodes
• AI Agents — 2× in this paper · explained in 2 episodes
• Alignment — 2× in this paper · explained in 1 episode
• Model Context Protocol — 2× in this paper
• Model Context Protocol (MCP) — 2× in this paper · explained in 8 episodes
• Robustness — 2× in this paper · explained in 5 episodes
• Adam — 1× in this paper · explained in 2 episodes
• Agent Harness — 1× in this paper · explained in 2 episodes
• Agent Interactions — 1× in this paper · explained in 1 episode
• Agentic AI — 1× in this paper · explained in 1 episode
• Consistency — 1× in this paper · explained in 3 episodes
• Failure Modes — 1× in this paper · explained in 3 episodes
• Fingerprinting — 1× in this paper · explained in 2 episodes
• GRPO — 1× in this paper · explained in 6 episodes
• Generative AI — 1× in this paper · explained in 2 episodes
• LLM-as-a-Judge — 1× in this paper · explained in 3 episodes
• Large Language Models (LLMs) — 1× in this paper · explained in 6 episodes
• Sparse Attention — 1× in this paper · explained in 5 episodes
• Two-Stage Framework — 1× in this paper · explained in 2 episodes

Sources

• SWE-chat: Coding Agent Interactions From Real Users in the Wild
• Seed2.0 Model Card: Towards Intelligence Frontier for Real-World Complexity
• DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence
• WildClawBench: A Benchmark for Real-World, Long-Horizon Agent Evaluation
• Seven simple steps for log analysis in AI systems
• Safety Testing LLM Agents at Scale: From Risk Discovery to Evidence-Grounded Verification
• GLM-5: from Vibe Coding to Agentic Engineering
• Seed1.5-VL Technical Report
• ProcCtrlBench: Evaluating Process-Level Defects and Control Preservation in LLM Coding Agents
• BenchTrace: A Benchmark for Testing Reflection Ability and Controlled Evolution in LLM Agents
• LURE: Live-Usage Replay Evaluations for Reducing Evaluation Awareness
• REAP: Automatic Curation of Coding Agent Benchmarks from Interactive Production Usage
• Kimi K3: Open Frontier Intelligence
• Evaluating whether AI models would sabotage AI safety research
• SHADE-Arena: Evaluating Sabotage and Monitoring in LLM Agents
• MiniMax Sparse Attention
• Who&When Pro: Can LLMs Really Attribute Failures in AI Agents?
• OpenClawBench: Benchmarking Process-side Anomalies in Real-world Agent Execution Trajectories
• RealClawBench: Live OpenClaw Benchmarks from Real Developer-Agent Sessions
• Insights Generator: Systematic Corpus-Level Trace Diagnostics for LLM Agents

Related papers

• MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
• Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
• Agents' Last Exam
• The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
• MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
• Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems