{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.28021", "kind": "paper", "url": "https://arxiv.org/abs/2608.28021v1"}, {"id": "gh:animeshshaw/geniac-secbench", "kind": "repo", "url": "https://github.com/AnimeshShaw/GenIaC-SecBench"}, {"id": "hf:animeshshaw/geniac-secbench", "kind": "dataset", "url": "https://huggingface.co/datasets/AnimeshShaw/GenIaC-SecBench"}], "authors": ["Animesh Shaw"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.28021", "source_url": "https://arxiv.org/abs/2608.28021"}, "categories": ["Cybersecurity", "Language & Knowledge", "cs.CR"], "description": {"en": "Evaluates LLM-generated Infrastructure-as-Code security using 100 scenarios across Terraform, CloudFormation, and Kubernetes, scanned by Checkov, Trivy, and KICS, with metrics including vulnerability density, schema validity, and LLM-judge review."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.28021v1", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.28021v1", "title": "GenIaC-SecBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2608.28021", "modality": null, "name": "GenIaC-SecBench", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "High", "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_geniac-secbench_393dfc82", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "0ec0c84b27e2a078043fbde2e696d82736a8be5db69ec0524799e4324669ff1e", "relation": "introduces", "review_model": "deepseek-v4-pro", "review_state": "ai-reviewed", "reviewed_at": "2026-09-05T08:28:38.611381Z", "source_url": "https://arxiv.org/abs/2608.28021v1"}, "publisher": null, "released": "2026-08-28", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.28021", "source_url": "https://arxiv.org/abs/2608.28021v1"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-28021", "source": "claire_radar", "source_benchmark_id": "2608.28021", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["Cybersecurity"], "area": "Language & Knowledge", "attention": {"asOf": "2026-09-05", "attemptedAt": "2026-09-05T01:35:52.848019Z", "githubRepo": "https://github.com/AnimeshShaw/GenIaC-SecBench", "githubScope": "benchmark_repo", "githubStars": 0, "hfDailySubmittedAt": null, "hfDataset": "AnimeshShaw/GenIaC-SecBench", "hfDatasetDownloads": 640, "hfDatasetLikes": 1, "hfPaperUpvotes": 0, "hfPaperUrl": "https://huggingface.co/papers/2608.28021", "observedAt": "2026-09-05T01:35:52.848019Z", "signalStatus": {"githubStars": {"observedAt": "2026-09-05T01:35:52.848019Z", "state": "fresh"}, "hfDatasetDownloads": {"observedAt": "2026-09-05T01:35:52.848019Z", "state": "fresh"}, "hfDatasetLikes": {"observedAt": "2026-09-05T01:35:52.848019Z", "state": "fresh"}, "hfPaperUpvotes": {"observedAt": "2026-09-05T01:35:52.848019Z", "state": "fresh"}}, "status": "fresh"}, "attentionForecast": {"confidence": "Medium", "horizon": "7d", "reason": "The benchmark addresses a timely practical concern in LLM-generated infrastructure security with clear artifacts and a human baseline, likely to attract moderate attention from the security and LLM evaluation communities.", "score": 62}, "capabilities": [], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-05T08:28:38.611381Z", "inputHash": "c733af17a7e13516d59165c408a0aecb6054d161d4d7c201176a823be6953c46", "model": "deepseek-v4-pro"}, "curation": {"canonicalNameEvidence": "# GenIaC-SecBench", "canonicalNameSource": "official_readme", "decisionReason": "The benchmark has a formal name, a clear repeatable evaluation protocol, explicit scoring metrics, and released code and data on GitHub and Hugging Face, supporting public comparison.", "model": "deepseek-v4-pro", "reviewedAt": "2026-09-05T08:28:38.611381Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "Evaluates LLM-generated Infrastructure-as-Code security using 100 scenarios across Terraform, CloudFormation, and Kubernetes, scanned by Checkov, Trivy, and KICS, with metrics including vulnerability density, schema validity, and LLM-judge review.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "score_submission", "evidence": {"reasonCodes": ["exact named benchmark artifact released in abstract", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL"], "snippet": "We introduce GenIaC-SecBench, a benchmark of 100 deployment scenarios stratified by architectural complexity, evaluated across 12 model configurations from four vendors, producing 1,196 IaC artifacts scanned by three independent policy engines (Checkov, Trivy, KICS)."}, "familyId": "bmf_3e1ff571a6a3", "firstSeenAt": "2026-08-31", "heat": null, "id": "bm_geniac-secbench_393dfc82", "indexedAt": "2026-08-31T07:22:53.604319Z", "industrySectors": ["Cybersecurity"], "links": {"code": "https://github.com/AnimeshShaw/GenIaC-SecBench", "data": "https://huggingface.co/datasets/AnimeshShaw/GenIaC-SecBench", "hfPaper": "https://huggingface.co/papers/2608.28021", "paper": "http://arxiv.org/abs/2608.28021v1", "pdf": "https://arxiv.org/pdf/2608.28021v1", "project": null, "report": "http://arxiv.org/abs/2608.28021v1"}, "metrics": [], "motivation": "Large language models are increasingly used to author Infrastructure-as-Code (IaC), where a single insecure default can be deployed directly into production.", "name": "GenIaC-SecBench", "oneLine": "Evaluates LLM-generated Infrastructure-as-Code security using 100 scenarios across Terraform, CloudFormation, and Kubernetes, scanned by Checkov, Trivy, and KICS, with metrics including vulnerability density, schema validity, and LLM-judge review.", "paperTitle": "Compared to What? A Human-Anchored Security Benchmark for LLM-Generated Infrastructure-as-Code", "primaryDomain": "Cybersecurity", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "http://arxiv.org/abs/2608.28021v1", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-09-05T06:02:50.019181Z"}, "publications": [], "ranking": {"30d": {"confidence": "High", "coverage": 1.0, "datasetDownloadRank": 9, "datasetRankPopulation": 38, "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": 0.0, "value": 0}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": 0.6634934098655181, "value": 640}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": 0.0, "value": 0}}, "confidence": "High", "coverage": 1.0, "rank": 209, "score": 10}, "method": "attention-ranking-v13", "rank": 209, "score": 10}, "90d": {"confidence": "High", "coverage": 1.0, "datasetDownloadRank": 22, "datasetRankPopulation": 73, "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": 0.0, "value": 0}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": 0.5363648885703535, "value": 640}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": 0.0, "value": 0}}, "confidence": "High", "coverage": 1.0, "rank": 431, "score": 16}, "method": "attention-ranking-v13", "rank": 431, "score": 16}}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-08-28", "source": {"authors": ["Animesh Shaw"], "categories": ["cs.CR", "cs.AI", "cs.MA", "cs.SE"], "id": "2608.28021", "title": "Compared to What? A Human-Anchored Security Benchmark for LLM-Generated Infrastructure-as-Code", "type": "arxiv", "url": "http://arxiv.org/abs/2608.28021v1"}, "sourceUpdatedAt": "2026-08-28T07:34:57Z", "topics": ["cs.CR"], "venueAttempts": [], "whyItMatters": "Provides a human-anchored baseline for IaC security evaluation, revealing that size-matched vulnerability density comparisons are essential to avoid confounding artifact size with security quality."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
