{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.23635", "kind": "paper", "url": "https://arxiv.org/abs/2608.23635"}], "authors": ["YiShan Zheng", "Yuan Wu", "Yi Chang"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.23635", "source_url": "https://arxiv.org/abs/2608.23635"}, "categories": ["Agents", "Code & Software", "General AI"], "description": {"en": "Evaluates tool-calling agents under perturbations across the tool-use pipeline, attributing failures to selection, grounding, argument binding, and feedback handling."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.23635", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.23635", "title": "ToolRobustBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2608.23635", "modality": null, "name": "ToolRobustBench", "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_toolrobustbench_a4d79725", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "ba54866aafa3005a7379c6f3d9c7f622574590edfe027caa2d2063035e8a5441", "relation": "introduces", "review_model": "deepseek-v4-pro", "review_state": "ai-reviewed", "reviewed_at": "2026-08-26T06:10:57.123445Z", "source_url": "https://arxiv.org/abs/2608.23635"}, "publisher": null, "released": "2026-08-23", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.23635", "source_url": "https://arxiv.org/abs/2608.23635"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-23635", "source": "claire_radar", "source_benchmark_id": "2608.23635", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Code & Software", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": null, "githubScope": null, "githubStars": null, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2608.23635", "observedAt": "2026-10-04T03:02:42.754940Z", "signalStatus": {"githubStars": {"state": "not_applicable"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "state": "unavailable"}}, "status": "unavailable"}, "attentionForecast": {"confidence": "Low", "horizon": "7d", "reason": "The benchmark targets a hot agent-tooling area with 15K instances across 7 models, but a supported public link is not supplied.", "score": 41}, "capabilities": [], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-26T06:10:57.123445Z", "inputHash": "e2f08dde5feb2e47a765dd6fe913e893119e3d487ff6fc4057f116f78faf9498", "model": "deepseek-v4-pro"}, "curation": {"canonicalNameEvidence": "We introduce ToolRobustBench, a stage-wise diagnostic benchmark for tool-calling agents", "canonicalNameSource": "abstract", "decisionReason": "The benchmark is formally named, introduces a repeatable evaluation with four perturbation families and scoring protocols, and describes 15,456 instances; paper implies public availability.", "model": "deepseek-v4-pro", "reviewedAt": "2026-08-26T06:10:57.123445Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "Evaluates tool-calling agents under perturbations across the tool-use pipeline, attributing failures to selection, grounding, argument binding, and feedback handling.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "score_submission", "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "We introduce ToolRobustBench, a stage-wise diagnostic benchmark for tool-calling agents, where a tool-calling agent is an LLM system that selects a tool, supplies structured arguments, and interprets its returned feedback."}, "familyId": "bmf_6c40e4d9d25b", "firstSeenAt": "2026-08-26", "heat": null, "id": "bm_toolrobustbench_a4d79725", "indexedAt": "2026-08-26T06:08:22.244942Z", "industrySectors": [], "links": {"code": null, "data": null, "paper": "https://arxiv.org/abs/2608.23635", "pdf": "https://arxiv.org/pdf/2608.23635", "project": null, "report": "https://arxiv.org/abs/2608.23635"}, "metrics": [], "motivation": "Large language models (LLMs) rely on tool calling as a fundamental agent capability, enabling them to invoke external systems and complete tasks beyond text generation.", "name": "ToolRobustBench", "oneLine": "Evaluates tool-calling agents under perturbations across the tool-use pipeline, attributing failures to selection, grounding, argument binding, and feedback handling.", "paperTitle": "ToolRobustBench: Stage-Wise Perturbation Evaluation and Failure Diagnosis for Tool-Calling Agents", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2608.23635", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-08-26T06:08:22.244942Z"}, "publications": [], "ranking": {"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": 1.0, "relation": "introduces", "releasedAt": "2026-08-23", "source": {"authors": ["YiShan Zheng", "Yuan Wu", "Yi Chang"], "categories": ["cs.SE", "cs.AI"], "id": "2608.23635", "title": "ToolRobustBench: Stage-Wise Perturbation Evaluation and Failure Diagnosis for Tool-Calling Agents", "type": "arxiv", "url": "https://arxiv.org/abs/2608.23635"}, "sourceUpdatedAt": "2026-08-23T15:22:49Z", "topics": ["Agents"], "venueAttempts": [], "whyItMatters": "Provides deterministic, cascade-aware diagnosis of robustness beyond clean tool-calling accuracy."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
