{
  "schema_version": 1,
  "builder_plan": {
    "scenario": "An agent reads external documents or tool responses that may contain malicious instructions.",
    "change": "Replay attacks and legitimate lookalikes in an isolated copy of your agent workflow.",
    "baseline": "Your existing checks with the same tasks, permissions, and tool responses.",
    "steps": [
      "Locate the paper's implementation and check its license, model support, data needs, and hardware requirements. Stop here if you cannot obtain the required artifacts.",
      "Save a small set of representative product tasks and expected outcomes. Keep evaluation tasks out of any training or tuning, and choose quality and cost limits before running.",
      "Run the baseline and the proposed change on identical inputs. Keep other settings fixed and repeat paired runs to expose variation.",
      "Save per-task outputs, failures, timings, and all usage costs. Review regressions before expanding the test."
    ],
    "measure": "Successful attacks missed, legitimate actions blocked, detection delay, and review cost.",
    "decision": "Keep the change only if it meets your preselected quality and latency limits and improves the metric you care about after setup and operating costs. Otherwise retain the baseline. This is an exploration plan, not a validated recommendation."
  },
  "builder_review": null,
  "arxiv_id": "2610.06514",
  "title": "ANT: A Multi-Granularity Network Traffic Dataset and Benchmark for Agents Behavior Auditing",
  "summary": "ANT replaces ad hoc or missing ground truth for auditing LLM agent behavior from network traffic by providing 3,114 annotated execution episodes with task, scenario, and behavior-primitive labels across 276,417 flows. It is a dataset and benchmark rather than a deployable method, so adoption costs are limited to storage and the effort of training or evaluating traffic classifiers against its 13 baselines. Its own results show that current traffic-analysis methods fail to identify risk when malicious workflows resemble benign tasks and cannot distinguish scenarios or rare behavior primitives with similar traffic patterns, so production auditing built on it would inherit those reliability gaps.",
  "rationale": "The paper contributes evaluation infrastructure, not a production technique, and its findings indicate existing traffic-based auditing methods are not yet reliable enough for deployment without further research.",
  "technique": "Network traffic dataset for agent behavior auditing",
  "family": "seguranca_e_guardrails",
  "recommendation": "observar",
  "published": "2026-10-05",
  "as_of": "2026-10-18",
  "editorial_status": "indexed",
  "source_url": "https://arxiv.org/abs/2610.06514",
  "independent_implementations": 1,
  "total_implementations": 2,
  "stars": 131,
  "citations": null,
  "claims": [
    {
      "claim_id": "claim-01",
      "statement": "ANT replaces ad hoc or missing ground truth for auditing LLM agent behavior from network traffic by providing 3,114 annotated execution episodes with task, scenario, and behavior-primitive labels across 276,417 flows. It is a dataset and benchmark rather than a deployable method, so adoption costs are limited to storage and the effort of training or evaluating traffic classifiers against its 13 baselines. Its own results show that current traffic-analysis methods fail to identify risk when malicious workflows resemble benign tasks and cannot distinguish scenarios or rare behavior primitives with similar traffic patterns, so production auditing built on it would inherit those reliability gaps.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2610.06514",
      "source_page": null,
      "source_excerpt": ""
    }
  ],
  "exposure_map": [
    {
      "dimension": "quality",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "compute",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "latency",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "operations",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "compatibility",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "security",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "data_and_training",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "dimension": "reproducibility",
      "basis": "not_evaluated",
      "finding": "",
      "interpretation": "",
      "limitation": "",
      "next_check": "",
      "source_url": "",
      "source_page": null,
      "source_excerpt": ""
    }
  ],
  "risks": [],
  "minimum_test": [],
  "open_questions": [],
  "independent_tests": [],
  "lab_reproductions": [],
  "report_available": false,
  "validation_tier": "unknown",
  "evidence_tier": "unknown",
  "infrastructure_basis": "unknown",
  "equations": [],
  "equations_status": "not_fetched",
  "core_kind": "",
  "equations_fetched_at": "",
  "equations_source_url": "",
  "equations_source_sha256": "",
  "exposures_reviewed_at": "",
  "exposures_source_sha256": ""
}