agent/papers/items/2026-2605-27690-traces-proactive-safety-auditing-for-multi-turn-llm-agents-via-trajectory-state-.md

Paper: TRACES: Proactive Safety Auditing for Multi-Turn LLM Agents via Trajectory-State Modeling

---

type: paper title: "TRACES: Proactive Safety Auditing for Multi-Turn LLM Agents via Trajectory-State Modeling" authors: Jiaqian Li, Yanshu Li, Boxuan Zhang, Ruixiang Tang, Kuan-Hao Huang year: 2026 venue: arXiv url: https://arxiv.org/abs/2605.27690 code_url: source: arxiv collected_at: 2026-07-08 published_at: 2026-05-26 updated_at: 2026-05-26 status: queued relevance: high topics:

• agent-evaluation
• agent-safety
• computer-use
• planning
• tool-use methods:

benchmarks:

models:

datasets:

• cs.CL
• cs.LG related_concepts:

related_jobs:

related_experiments:

related_projects:

collection_score: 18 collection_queries: agent-safety

One-line Takeaway

Auto-collected from arXiv because it matched the Agent collection queries. Needs human skim.

Why Collected

• matched queries: agent-safety
• inferred topics: agent-evaluation, agent-safety, computer-use, planning, tool-use
• arXiv categories: cs.CL, cs.LG
• collection score: 18

Review Checklist

• Does this paper directly inform Agent architecture, evaluation, memory, tools, safety, coding agents, GUI/browser agents, or multi-agent workflows?
• Does it include a benchmark, dataset, code, or reproducible experimental setup?
• Should it be promoted from queued to skimmed or summarized?

Links

• arXiv: https://arxiv.org/abs/2605.27690