{"record": {"aliases": ["agentic-security-benchmark"], "artifacts": [{"id": "gh:gitayg/agentic-security-benchmark", "kind": "repo", "url": "https://github.com/gitayg/agentic-security-benchmark"}, {"kind": "website", "url": "https://glick.run/testing-methodology.html"}], "categories": ["Language & Knowledge", "Software & AI Compute", "cs.AI"], "description": {"en": "Measures detection and prevention of agentic-AI security products on 286 attack and 875 benign samples across five AMTSO attack vectors via a scoring adapter harness."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://github.com/gitayg/agentic-security-benchmark", "source": "claire_radar", "source_url": "https://github.com/gitayg/agentic-security-benchmark", "title": "Agentic Security Benchmark"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:github:gitayg/agentic-security-benchmark", "modality": null, "name": "Agentic Security Benchmark", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"admission_policy_version": "2026-09-09.1", "confidence": "Medium", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-candidate", "display_eligible": "true", "origin_record_id": "bm_agentic-security-benchmark_9165db66", "origin_source": "github", "recognition_confidence": "0.4", "record_sha256": "1080a7bc267829aeff2fb3b866204b6d13678f35a473e4ee47ecb0ffdf0e00f2", "relation": "unclear", "review_model": "deepseek-v4-pro", "review_state": "ai-reviewed", "reviewed_at": "2026-09-11T06:20:43.215764Z", "source_url": "https://github.com/gitayg/agentic-security-benchmark"}, "publisher": null, "released": "2026-09-07", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:github:gitayg/agentic-security-benchmark", "source_url": "https://github.com/gitayg/agentic-security-benchmark"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-github-gitayg-agentic-security-benchmark", "source": "claire_radar", "source_benchmark_id": "github:gitayg/agentic-security-benchmark", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": ["agentic-security-benchmark"], "annotation": "Unknown", "applicationDomains": ["Software & AI Compute"], "area": "Language & Knowledge", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/gitayg/agentic-security-benchmark", "githubScope": "benchmark_repo", "githubStars": 0, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": null, "observedAt": "2026-10-04T15:31:48.030916Z", "signalStatus": {"githubStars": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"state": "not_applicable"}}, "status": "fresh"}, "attentionForecast": {"confidence": "Low", "horizon": "7d", "reason": "Niche benchmark with limited public discoverability and no mention in supplied social signals, but a complete reused data path and security practitioner interest may drive moderate attention.", "score": 35}, "capabilities": [], "confidence": "Medium", "construction": "Transform Existing", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-11T06:20:43.215764Z", "inputHash": "46449d6d5db313de9d257e2dc405c0bf3b647c095653d626ab3d6a7530ff17ef", "model": "deepseek-v4-pro"}, "curation": {"admissionPolicyVersion": "2026-09-09.1", "canonicalNameEvidence": "# Agentic Security Benchmark **A corpus and a harness for measuring what an agentic-AI security product actually stops.**", "canonicalNameSource": "official_readme", "decisionReason": "The repository provides a named benchmark, corpus, scoring harness, and clear instructions for submission, meeting publish criteria.", "evaluationEvidence": {"reuse": "```bash git clone && cd agentic-security-benchmark npm run score -- --adapter keyword # no npm install — there is nothing to install ```", "scoring": "You get a recall number, a false-positive number, and a distribution of outcomes graded against [AMTSO's *Guidelines for Testing of Agentic Security Products* v1.0]", "task": "**286 attack samples, 875 benign**, across five AMTSO attack vectors."}, "model": "deepseek-v4-pro", "reviewedAt": "2026-09-11T06:20:43.215764Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-candidate", "demo": false, "description": "Measures detection and prevention of agentic-AI security products on 286 attack and 875 benign samples across five AMTSO attack vectors via a scoring adapter harness.", "discoverySources": [{"id": "github:gitayg/agentic-security-benchmark", "type": "github", "url": "https://github.com/gitayg/agentic-security-benchmark"}], "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "score_submission", "evidence": {"reasonCodes": ["discovered via github", "benchmark term in abstract", "evaluation protocol evidence", "public artifact URL", "no explicit benchmark release evidence"], "snippet": "agentic-security-benchmark A corpus and harness for measuring what an agentic-AI security product actually stops."}, "familyId": "bmf_cc6aa0872aa2", "firstSeenAt": "2026-09-08", "heat": null, "id": "bm_agentic-security-benchmark_9165db66", "indexedAt": "2026-09-08T06:12:01.862698Z", "industrySectors": ["Software & Cloud"], "links": {"code": "https://github.com/gitayg/agentic-security-benchmark", "project": "https://glick.run/testing-methodology.html"}, "metrics": [], "motivation": "agentic-security-benchmark A corpus and harness for measuring what an agentic-AI security product actually stops.", "name": "Agentic Security Benchmark", "oneLine": "Measures detection and prevention of agentic-AI security products on 286 attack and 875 benign samples across five AMTSO attack vectors via a scoring adapter harness.", "paperTitle": "agentic-security-benchmark", "primaryDomain": "Software & AI Compute", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://github.com/gitayg/agentic-security-benchmark", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-09-08T06:12:01.862698Z"}, "publications": [], "publishers": [{"name": "gitayg", "organizationType": "community", "role": "benchmark-publisher", "sourceUrl": "https://github.com/gitayg/agentic-security-benchmark"}], "ranking": {"30d": {"confidence": "Low", "coverage": 0.55, "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": 30, "historyDate": "2026-09-05", "rank": null, "score": null, "windowDays": 30}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": 0.0, "value": 0}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": null, "value": null}}, "confidence": "Low", "coverage": 0.55, "rank": 201, "score": 23}, "method": "attention-ranking-v13", "rank": 201, "score": 23}, "90d": {"confidence": "Low", "coverage": 0.55, "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": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": null, "value": null}}, "confidence": "Low", "coverage": 0.55, "rank": 465, "score": 23}, "method": "attention-ranking-v13", "rank": 465, "score": 23}}, "readiness": "Runnable", "recognitionConfidence": 0.4, "relation": "unclear", "releasedAt": "2026-09-07", "source": {"authors": ["gitayg"], "categories": [], "id": "github:gitayg/agentic-security-benchmark", "publicSignals": {"githubStars": 0}, "title": "agentic-security-benchmark", "type": "github", "url": "https://github.com/gitayg/agentic-security-benchmark"}, "sourceUpdatedAt": "2026-09-08T02:18:26Z", "topics": ["cs.AI"], "venueAttempts": [], "whyItMatters": "Provides a shared, reproducible way to compare agentic security product performance, exposing false positives and marginal contribution over model refusal baselines."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
