{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.22808", "kind": "paper", "url": "https://arxiv.org/abs/2608.22808v1"}, {"id": "gh:yzhao062/catchbench", "kind": "repo", "url": "https://github.com/yzhao062/catchbench"}], "authors": ["Yue Zhao", "Mengyuan Li", "Ruolin Li", "Prince Zizhuang Wang", "Shuli Jiang", "Linsey Pang", "Xiongye Xiao", "Xiyang Hu"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.22808", "source_url": "https://arxiv.org/abs/2608.22808"}, "categories": ["Agents", "General AI", "Language & Knowledge"], "description": {"en": "Assesses agent failure auditing across PRE, LIVE, and POST information states with seven task contracts covering evidential and Gold-derived diagnostics."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.22808v1", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.22808v1", "title": "CatchBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["University of Southern California"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2608.22808", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2608.22808", "source_url": "https://openalex.org/W7204175720"}, "key": "claire-radar:2608.22808", "modality": null, "name": "CatchBench", "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_catchbench_f09872b4", "origin_source": "arxiv", "recognition_confidence": "0.75", "record_sha256": "67d669911d3bd41236ad0f3dd213c3dfd2194c15e9481efa2c3ee488a908f395", "relation": "introduces", "review_model": "deepseek-v4-pro", "review_state": "ai-reviewed", "reviewed_at": "2026-08-27T04:18:56.318436Z", "source_url": "https://arxiv.org/abs/2608.22808v1"}, "publisher": null, "released": "2026-08-24", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.22808", "source_url": "https://arxiv.org/abs/2608.22808v1"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-22808", "source": "claire_radar", "source_benchmark_id": "2608.22808", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Language & Knowledge", "attention": {"asOf": "2026-08-26", "attemptedAt": "2026-08-25T16:48:55.714492Z", "githubRepo": "https://github.com/yzhao062/catchbench", "githubScope": "benchmark_repo", "githubStars": 4, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2608.22808", "observedAt": "2026-08-25T16:48:55.714492Z", "signalStatus": {"githubStars": {"observedAt": "2026-08-25T16:48:55.714492Z", "state": "fresh"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-08-25T16:48:55.714492Z", "state": "unavailable"}}, "status": "fresh"}, "attentionForecast": {"confidence": "Medium", "horizon": "7d", "reason": "Broad agent-auditing scope and a detailed released repository suggest moderate initial visibility, though the niche multi-state framing may limit rapid spread.", "score": 55}, "capabilities": [], "confidence": "Medium", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-27T04:18:56.318436Z", "inputHash": "fdae69c8d88d3ee5e1705f7dff2dbab798cc9177df8a99c01f974b5efe1b3b27", "model": "deepseek-v4-pro"}, "curation": {"canonicalNameEvidence": "CatchBench therefore puts one auditor's question to three information states", "canonicalNameSource": "abstract", "decisionReason": "Formally named benchmark with released code and data, seven task contracts with defined metrics, and a clear public reuse path via GitHub.", "model": "deepseek-v4-pro", "reviewedAt": "2026-08-27T04:18:56.318436Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "Assesses agent failure auditing across PRE, LIVE, and POST information states with seven task contracts covering evidential and Gold-derived diagnostics.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "evaluation protocol evidence", "public artifact URL"], "snippet": "A benchmark number is therefore not interpretable until the process behind its labels is published and tested for the shortcut it may leave."}, "familyId": "bmf_4e634670acb6", "firstSeenAt": "2026-08-25", "heat": null, "id": "bm_catchbench_f09872b4", "indexedAt": "2026-08-25T16:21:05.134052Z", "industrySectors": [], "links": {"code": "https://github.com/yzhao062/catchbench", "data": null, "paper": "http://arxiv.org/abs/2608.22808v1", "pdf": "https://arxiv.org/pdf/2608.22808v1", "project": null, "report": "http://arxiv.org/abs/2608.22808v1"}, "metrics": [], "motivation": "When can an agent failure be caught?", "name": "CatchBench", "oneLine": "Assesses agent failure auditing across PRE, LIVE, and POST information states with seven task contracts covering evidential and Gold-derived diagnostics.", "paperTitle": "CatchBench: When Can an Agent Failure Be Caught?", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "http://arxiv.org/abs/2608.22808v1", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-08-25T16:21:05.134052Z"}, "publications": [], "publishers": [{"name": "CatchBench maintainers", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://github.com/yzhao062/catchbench"}], "ranking": {"30d": {"confidence": "Low", "coverage": 0.55, "growth": {"components": {"githubStars": {"percentile": null, "value": null}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 30}, "level": {"components": {"githubStars": {"percentile": 0.6766917293233082, "value": 4}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.55, "rank": 45, "score": 60}, "method": "attention-ranking-v11", "rank": 45, "score": 60}, "90d": {"confidence": "Low", "coverage": 0.55, "growth": {"components": {"githubStars": {"percentile": null, "value": null}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 90}, "level": {"components": {"githubStars": {"percentile": 0.4832869080779944, "value": 4}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.55, "rank": 199, "score": 49}, "method": "attention-ranking-v11", "rank": 199, "score": 49}, "today": {"confidence": "Medium", "coverage": 0.7, "growth": {"components": {"githubStars": {"percentile": null, "value": null}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}, "llmAttentionForecast": {"percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": 2, "historyDate": "2026-08-24", "rank": null, "score": null, "windowDays": 1}, "level": {"components": {"githubStars": {"percentile": 0.5, "value": 4}, "hfDatasetDownloads": {"percentile": null, "value": null}, "hfPaperUpvotes": {"percentile": null, "value": null}, "llmAttentionForecast": {"percentile": 0.6, "value": 60}}, "confidence": "Medium", "coverage": 0.7, "rank": 3, "score": 54}, "method": "attention-ranking-v11", "rank": 3, "score": 54}}, "readiness": "Runnable", "recognitionConfidence": 0.75, "relation": "introduces", "releasedAt": "2026-08-24", "source": {"authors": ["Yue Zhao"], "categories": ["cs.LG", "cs.MA", "cs.PF"], "id": "2608.22808", "title": "CatchBench: When Can an Agent Failure Be Caught?", "type": "arxiv", "url": "http://arxiv.org/abs/2608.22808v1"}, "sourceUpdatedAt": "2026-08-24T05:09:41Z", "topics": ["Agents"], "venueAttempts": [], "whyItMatters": "Unifies previously separate auditing settings under one interface and exposes label-process shortcuts, helping users judge whether benchmark scores reflect reasoning."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
