{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.26882", "kind": "paper", "url": "https://arxiv.org/abs/2608.26882"}], "authors": ["Yitian Zhou", "Jingyu Zheng", "Qiliang Jiang", "Linkang Du", "Haoming Liu", "Lichao Wu", "Shiyi Zhao", "Mengxiang Liu", "Ruilong Deng"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.26882", "source_url": "https://arxiv.org/abs/2608.26882"}, "categories": ["Agents & Tool Use", "Factuality", "Manufacturing & Process Control", "Robotics & Embodied AI", "cs.CR"], "description": {"en": "Evaluates autonomous LLM agents on real programmable logic controllers connected to closed-loop process simulations, measuring their ability to achieve sustained physical impact through fixed diagnostic flags applied to runner, communication, PLC-object, and process records."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.26882", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.26882", "title": "PLCBENCH"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Zhejiang University", "Xi'an Jiaotong University", "University of Bristol"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2608.26882", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2608.26882", "source_url": "https://openalex.org/W7204604947"}, "key": "claire-radar:2608.26882", "modality": null, "name": "PLCBENCH", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "Medium", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-indexed", "display_eligible": "true", "origin_record_id": "bm_plcbench_fb7005e5", "origin_source": "arxiv", "recognition_confidence": "0.65", "record_sha256": "ac15e6e37215e12a1096c25cf7f2af982fd33d96e958fb09b3e4512454f2dfe4", "relation": "introduces", "review_model": "deepseek-v4-pro", "review_state": "ai-reviewed", "reviewed_at": "2026-09-08T08:58:25.506689Z", "source_url": "https://arxiv.org/abs/2608.26882"}, "publisher": null, "released": "2026-08-27", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.26882", "source_url": "https://arxiv.org/abs/2608.26882"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-26882", "source": "claire_radar", "source_benchmark_id": "2608.26882", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["Manufacturing & Process Control", "Robotics & Embodied AI"], "area": "Agents & Tool Use", "attention": {"asOf": "2026-09-08", "attemptedAt": "2026-09-08T01:38:09.842216Z", "githubRepo": null, "githubScope": null, "githubStars": null, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2608.26882", "observedAt": "2026-09-07T01:28:43.988909Z", "signalStatus": {"githubStars": {"state": "not_applicable"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-09-08T01:38:09.842216Z", "state": "unavailable"}}, "status": "unavailable"}, "attentionForecast": {"confidence": "Medium", "horizon": "7d", "reason": "The first real-PLC hardware-in-the-loop benchmark for LLM agents in industrial control systems addresses a high-stakes security gap with detailed reproducibility support, likely attracting attention from security and AI evaluation communities.", "score": 62}, "capabilities": ["Factuality"], "confidence": "Medium", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-08T08:58:25.506689Z", "inputHash": "a1c8182e8222244f3532e8a188869186c9ada4254ee9c705162fb5c878be6d6a", "model": "deepseek-v4-pro"}, "curation": {"canonicalNameEvidence": "We present PLCBENCH, to our knowledge, the first real-PLC hardware-in-the-loop (HIL) framework for characterizing this cyber-to-physical capability and its boundaries.", "canonicalNameSource": "abstract", "decisionReason": "Formally named PLCBENCH with a deterministic evaluator, fixed scoring rules, and released code plus a software-only reproduction pipeline, enabling other teams to run and compare results.", "model": "deepseek-v4-pro", "reviewedAt": "2026-09-08T08:58:25.506689Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "Evaluates autonomous LLM agents on real programmable logic controllers connected to closed-loop process simulations, measuring their ability to achieve sustained physical impact through fixed diagnostic flags applied to runner, communication, PLC-object, and process records.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "score_submission", "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "evaluation protocol evidence"], "snippet": "We present PLCBENCH, to our knowledge, the first real-PLC hardware-in-the-loop (HIL) framework for characterizing this cyber-to-physical capability and its boundaries."}, "familyId": "bmf_db5390e10777", "firstSeenAt": "2026-08-30", "heat": null, "id": "bm_plcbench_fb7005e5", "indexedAt": "2026-08-30T06:59:55.242506Z", "industrySectors": ["Manufacturing", "Robotics"], "links": {"code": null, "data": null, "paper": "https://arxiv.org/abs/2608.26882", "pdf": "https://arxiv.org/pdf/2608.26882", "project": null, "report": "https://arxiv.org/abs/2608.26882"}, "metrics": [], "motivation": "Industrial control systems (ICSs) rely on programmable logic controllers (PLCs) to connect networked computation with physical control.", "name": "PLCBENCH", "oneLine": "Evaluates autonomous LLM agents on real programmable logic controllers connected to closed-loop process simulations, measuring their ability to achieve sustained physical impact through fixed diagnostic flags applied to runner, communication, PLC-object, and process records.", "paperTitle": "PLCBench: Can Autonomous LLM Agents Turn PLC Access into Sustained Physical Impact?", "primaryDomain": "Manufacturing & Process Control", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2608.26882", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-08-30T06:59:55.242506Z"}, "publications": [], "ranking": {"30d": {"confidence": "Low", "coverage": 0.0, "growth": {"components": {"githubStars": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfDatasetDownloads": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfPaperUpvotes": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 30}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": null, "value": null}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "rank": null, "score": null}, "method": "attention-ranking-v13", "rank": null, "score": null}, "90d": {"confidence": "Low", "coverage": 0.0, "growth": {"components": {"githubStars": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfDatasetDownloads": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfPaperUpvotes": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 90}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": null, "value": null}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "rank": null, "score": null}, "method": "attention-ranking-v13", "rank": null, "score": null}}, "readiness": "Paper only", "recognitionConfidence": 0.65, "relation": "introduces", "releasedAt": "2026-08-27", "source": {"authors": ["Yitian Zhou", "Jingyu Zheng", "Qiliang Jiang", "Linkang Du", "Haoming Liu", "Lichao Wu", "Shiyi Zhao", "Mengxiang Liu", "and Ruilong Deng"], "categories": ["cs.CR", "cs.AI"], "id": "2608.26882", "title": "PLCBench: Can Autonomous LLM Agents Turn PLC Access into Sustained Physical Impact?", "type": "arxiv", "url": "https://arxiv.org/abs/2608.26882"}, "sourceUpdatedAt": "2026-08-27T09:40:11Z", "topics": ["cs.CR"], "venueAttempts": [], "whyItMatters": "Addresses the lack of evaluations that extend beyond digital tasks or software exploitation to sustained cyber-physical effects, providing a reproducible way to assess LLM-driven threats to industrial control systems and identify intervention points for defense."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
