{
  "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": "2610.31404",
  "title": "AgentPhishing: A Benchmark of LLM Email Agents Under Adaptive Attack",
  "summary": "AgentPhishing replaces reliance on conventional static phishing datasets for assessing LLM email agents, adding an optimizer (PhishOpt) that generates persuasive, agent-adapted attacks under realism constraints plus a reusable universal generation skill. The cost is additional evaluation infrastructure: an attack-optimization loop, API or GPU compute for the generator and target agents, and analysis effort beyond a standard detection benchmark. It can fail if the optimized attacks overfit to the tested agents or the realism constraint is weak, so detection and compliance rates may not transfer to real-world phishing or to agents with different tool configurations.",
  "rationale": "Any team shipping an email-handling agent should run adversarial compliance testing rather than trusting conventional phishing benchmarks, but the specific benchmark's predictive value requires validation against the team's own agent stack.",
  "technique": "Adaptive phishing benchmark for LLM email agents",
  "family": "seguranca_e_guardrails",
  "recommendation": "testar",
  "published": "2026-10-22",
  "as_of": "2026-10-23",
  "editorial_status": "indexed",
  "source_url": "https://arxiv.org/abs/2610.31404",
  "independent_implementations": 1,
  "total_implementations": 2,
  "stars": 10,
  "citations": null,
  "claims": [
    {
      "claim_id": "claim-01",
      "statement": "AgentPhishing replaces reliance on conventional static phishing datasets for assessing LLM email agents, adding an optimizer (PhishOpt) that generates persuasive, agent-adapted attacks under realism constraints plus a reusable universal generation skill. The cost is additional evaluation infrastructure: an attack-optimization loop, API or GPU compute for the generator and target agents, and analysis effort beyond a standard detection benchmark. It can fail if the optimized attacks overfit to the tested agents or the realism constraint is weak, so detection and compliance rates may not transfer to real-world phishing or to agents with different tool configurations.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2610.31404",
      "source_page": null,
      "source_excerpt": ""
    },
    {
      "claim_id": "claim-02",
      "statement": "In an evaluation of nine agents, four exhibit phishing-detection rates below 10% and malicious-request compliance rates above 45% under adaptive attacks, despite handling pre-existing phishing emails safely.",
      "basis": "inferred",
      "result": "",
      "baseline": "",
      "conditions": "",
      "source_url": "https://arxiv.org/abs/2610.31404",
      "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": ""
}