{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.09096", "kind": "paper", "url": "https://arxiv.org/abs/2608.09096"}, {"id": "gh:rucaibox/evo-bench", "kind": "repo", "url": "https://github.com/RUCAIBox/Evo-Bench"}], "authors": ["Lisheng Huang", "Chen Yang", "Hao Zhou", "Huatong Song", "Zongchao Chen", "Ran Le", "Yang Song", "Wayne Xin Zhao", "Tao Zhang"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.09096", "source_url": "https://arxiv.org/abs/2608.09096"}, "categories": ["Agents", "General AI", "Language & Knowledge"], "description": {"en": "Evo-Bench is a benchmark for evaluating whether language models can autonomously improve their agent harness, with 608 harness-sensitive tasks across Search, Office, and General domains, fixed policy model, and resource budget."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.09096", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.09096", "title": "Evo-Bench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Renmin University of China"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2608.09096", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2608.09096", "source_url": "https://openalex.org/W7202216241"}, "key": "claire-radar:2608.09096", "modality": null, "name": "Evo-Bench", "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": "unknown", "origin_record_id": "bm_evo-bench_a13a17e9", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "bd7c6140909b1b8773a0608641d3ab044104860aa1cf83406165a28b9bc0ba03", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2608.09096"}, "publisher": null, "released": "2026-08-10", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.09096", "source_url": "https://arxiv.org/abs/2608.09096"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-09096", "source": "claire_radar", "source_benchmark_id": "2608.09096", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Language & Knowledge", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/RUCAIBox/Evo-Bench", "githubScope": "benchmark_repo", "githubStars": 19, "hfDailySubmittedAt": "2026-08-11T00:00:00.000Z", "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 18, "hfPaperUrl": "https://huggingface.co/papers/2608.09096", "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": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}}, "status": "fresh"}, "capabilities": [], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "84ac3298bdc134f6072c77d653b047219d85d1f68801c26089d65708bb1e5d62", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "Evo-Bench is a benchmark for evaluating whether language models can autonomously improve their agent harness, with 608 harness-sensitive tasks across Search, Office, and General domains, fixed policy model, and resource budget.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "To address these challenges, we introduce Evo-Bench, the first benchmark designed to evaluate models' intrinsic harness-evolving capabilities across Search, Office, and General agent domains."}, "familyId": "bmf_d555691c91f5", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [{"evidence": [{"basis": "existing-code-link", "linkedUrl": "https://github.com/RUCAIBox/Evo-Bench", "url": "https://github.com/RUCAIBox/Evo-Bench"}, {"basis": "existing-metric-association", "linkedUrl": "https://github.com/RUCAIBox/Evo-Bench", "url": "https://huggingface.co/papers/2608.09096"}, {"basis": "huggingface-paper-association", "linkedUrl": "https://github.com/RUCAIBox/Evo-Bench", "url": "https://huggingface.co/api/papers/2608.09096"}], "scope": "benchmark_repo", "url": "https://github.com/RUCAIBox/Evo-Bench", "verification": "source-linked"}], "status": "linked"}, "heat": null, "id": "bm_evo-bench_a13a17e9", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/RUCAIBox/Evo-Bench", "data": null, "hfPaper": "https://huggingface.co/papers/2608.09096", "paper": "https://arxiv.org/abs/2608.09096", "pdf": "https://arxiv.org/pdf/2608.09096", "project": null, "report": "https://arxiv.org/abs/2608.09096"}, "metrics": [], "motivation": "Large Language Models (LLMs) have driven rapid progress in autonomous agents, yet standard evaluations remain confined to static task solving.", "name": "Evo-Bench", "oneLine": "Evo-Bench is a benchmark for evaluating whether language models can autonomously improve their agent harness, with 608 harness-sensitive tasks across Search, Office, and General domains, fixed policy model, and resource budget.", "paperTitle": "Evo-Bench: Can Language Models Improve Agent Harness?", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2608.09096", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "RUCAIBox", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://github.com/RUCAIBox/Evo-Bench"}], "ranking": {"90d": {"confidence": "Medium", "coverage": 0.7, "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": 0.3998764068733198, "value": 19}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": 0.5202664955757326, "value": 18}}, "confidence": "Medium", "coverage": 0.7, "rank": 106, "score": 45}, "method": "attention-ranking-v13", "rank": 106, "score": 45}}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-08-10", "source": {"authors": ["Lisheng Huang", "Chen Yang", "Hao Zhou", "Huatong Song", "Zongchao Chen", "Ran Le", "Yang Song", "Wayne Xin Zhao", "Tao Zhang"], "categories": ["cs.CL"], "id": "2608.09096", "title": "Evo-Bench: Can Language Models Improve Agent Harness?", "type": "arxiv", "url": "https://arxiv.org/abs/2608.09096"}, "sourceUpdatedAt": "2026-08-11T02:53:55Z", "topics": ["Agents"], "venueAttempts": [], "whyItMatters": "It isolates harness evolution from base model strength, addressing a gap in evaluating agent self-improvement capabilities."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
