{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.07297", "kind": "paper", "url": "https://arxiv.org/abs/2606.07297"}, {"id": "gh:qiushao-e/swe-explore-bench", "kind": "repo", "url": "https://github.com/Qiushao-E/SWE-Explore-Bench"}], "authors": ["Shaoqiu Zhang", "Yuhang Wang", "Jialiang Liang", "Yuling Shi", "Wenhao Zeng", "Maoquan Wang", "Shilin He", "Ningyuan Xu", "Siyu Ye", "Kai Cai", "Xiaodong Gu"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.07297", "source_url": "https://arxiv.org/abs/2606.07297"}, "categories": ["Code & Software", "Software & AI Compute", "cs.SE"], "description": {"en": "SWE-Explore evaluates repository exploration by coding agents: given a repository and an issue, an explorer returns a ranked list of relevant code regions under a fixed line budget. Ground truth is line-level, derived from successful repair trajectories. Coverage, ranking, and context-efficiency metrics are scored."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.07297", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.07297", "title": "SWE-Explore"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Shanghai Jiao Tong University", "Xinjiang University", "Chinese University of Hong Kong", "University of Illinois Urbana-Champaign"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2606.07297", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2606.07297", "source_url": "https://openalex.org/W7163915557"}, "key": "claire-radar:2606.07297", "modality": null, "name": "SWE-Explore", "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_swe-explore_ece5d09b", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "8f9f304cf9b2a276e8ab1cc2da5db47a0a14f7b9603a06ed3b9fc79f56d70580", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.07297"}, "publisher": null, "released": "2026-06-05", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.07297", "source_url": "https://arxiv.org/abs/2606.07297"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-07297", "source": "claire_radar", "source_benchmark_id": "2606.07297", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["Software & AI Compute"], "area": "Code & Software", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/Qiushao-E/SWE-Explore-Bench", "githubScope": "benchmark_repo", "githubStars": 43, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 123, "hfPaperUrl": "https://huggingface.co/papers/2606.07297", "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": {"checkedAt": "2026-10-04T15:31:48.030916Z", "lastSuccessfulDate": "2026-09-25", "state": "stale"}}, "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": "1648e19fc75788d56e640247218b1b7764bbe77aae0bea5d3e5143dfb337fcc2", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "SWE-Explore evaluates repository exploration by coding agents: given a repository and an issue, an explorer returns a ranked list of relevant code regions under a fixed line budget. Ground truth is line-level, derived from successful repair trajectories. Coverage, ranking, and context-efficiency metrics are scored.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact coined title identity tied to benchmark evidence", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "In this paper, we introduce SWE-Explore, a benchmark that isolates the evaluation of repository exploration, a critical capability of coding agents."}, "familyId": "bmf_1b67c9f91522", "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/Qiushao-E/SWE-Explore-Bench", "url": "https://github.com/Qiushao-E/SWE-Explore-Bench"}, {"basis": "existing-metric-association", "linkedUrl": "https://github.com/Qiushao-E/SWE-Explore-Bench", "url": "https://huggingface.co/papers/2606.07297"}], "scope": "benchmark_repo", "url": "https://github.com/Qiushao-E/SWE-Explore-Bench", "verification": "source-linked"}], "status": "linked"}, "heat": null, "id": "bm_swe-explore_ece5d09b", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": ["Software & Cloud"], "links": {"code": "https://github.com/Qiushao-E/SWE-Explore-Bench", "data": null, "hfPaper": "https://huggingface.co/papers/2606.07297", "paper": "https://arxiv.org/abs/2606.07297", "pdf": "https://arxiv.org/pdf/2606.07297", "project": null, "report": "https://arxiv.org/abs/2606.07297"}, "metrics": [], "motivation": "Repository-level coding benchmarks such as SWE-bench have driven a rapid surge in the capabilities of coding agents.", "name": "SWE-Explore", "oneLine": "SWE-Explore evaluates repository exploration by coding agents: given a repository and an issue, an explorer returns a ranked list of relevant code regions under a fixed line budget. Ground truth is line-level, derived from successful repair trajectories. Coverage, ranking, and context-efficiency metrics are scored.", "paperTitle": "SWE-Explore: Benchmarking How Coding Agents Explore Repositories", "primaryDomain": "Software & AI Compute", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2606.07297", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "ranking": {}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-06-05", "source": {"authors": ["Shaoqiu Zhang", "Yuhang Wang", "Jialiang Liang", "Yuling Shi", "Wenhao Zeng", "Maoquan Wang", "Shilin He", "Ningyuan Xu", "Siyu Ye", "Kai Cai", "Xiaodong Gu"], "categories": ["cs.SE", "cs.CL"], "id": "2606.07297", "title": "SWE-Explore: Benchmarking How Coding Agents Explore Repositories", "type": "arxiv", "url": "https://arxiv.org/abs/2606.07297"}, "sourceUpdatedAt": "2026-06-05T14:08:27Z", "topics": ["cs.SE"], "venueAttempts": [], "whyItMatters": "Existing repository-level benchmarks treat coding tasks as a single resolved/unresolved outcome, obscuring whether an agent locates the right context. SWE-Explore isolates exploration quality, enabling targeted evaluation of retrieval and localization capabilities that precede patch generation. Its metrics track downstream repair behavior, offering practical value for improving agent design."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
