{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2609.12345", "kind": "paper", "url": "https://arxiv.org/abs/2609.12345"}, {"id": "gh:gbw206/pararecover", "kind": "repo", "url": "https://github.com/gbw206/ParaRecover"}], "authors": ["Bowen Guan", "Zhentao Yin", "Yanming Shen"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2609.12345", "source_url": "https://arxiv.org/abs/2609.12345"}, "categories": ["General AI", "Language & Knowledge", "cs.LG"], "description": {"en": "ParaRecover evaluates LLM-based parallel tool-use agents on error localization and recovery, testing whether agents detect execution errors, diagnose root causes, and repair plans when tool calls fail. It scores replanner output across structural, diagnostic, and evolutionary dimensions."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2609.12345", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2609.12345", "title": "ParaRecover"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2609.12345", "modality": null, "name": "ParaRecover", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"admission_policy_version": "2026-09-09.1", "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_pararecover_c43c0d4d", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "e9d98811800d7da8cccee1891c199636b8552d1fae34b58f9b67e5e7dff6d41a", "relation": "introduces", "review_model": "deepseek-v4-pro", "review_state": "ai-reviewed", "reviewed_at": "2026-09-14T00:43:53.625753Z", "source_url": "https://arxiv.org/abs/2609.12345"}, "publisher": null, "released": "2026-09-11", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2609.12345", "source_url": "https://arxiv.org/abs/2609.12345"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2609-12345", "source": "claire_radar", "source_benchmark_id": "2609.12345", "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/gbw206/ParaRecover", "githubScope": "benchmark_repo", "githubStars": 0, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 0, "hfPaperUrl": "https://huggingface.co/papers/2609.12345", "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"}, "attentionForecast": {"confidence": "Low", "horizon": "7d", "reason": "Agent reliability and tool-use recovery are actively discussed topics, but the record shows a focused task scope and no external adoption or popularity signals.", "reason_note": "", "score": 42}, "capabilities": [], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-14T10:07:09.772218Z", "inputHash": "a2da4ac0e9499b7d955795bb2112faf8b99c2f57ef6d19e29bdbb3206d87d6e2", "model": "deepseek-v4-flash"}, "curation": {"admissionPolicyVersion": "2026-09-09.1", "canonicalNameEvidence": "ParaRecover: A Process-Level Benchmark for Error Localization and Recovery in Parallel Tool-Use Agents", "canonicalNameSource": "paper_title", "decisionReason": "Code and data are available with clear execution scripts and defined evaluation dimensions, enabling other teams to run and extend the benchmark.", "evaluationEvidence": {"reuse": "python scripts/run_batch_replanner.py --input data/LEVEL-1/LEVEL-1_augmented.jsonl", "scoring": "The benchmark evaluates replanner output across three dimensions (14 sub-items + 1 script check).", "task": "It tests whether an agent can detect execution errors, diagnose root causes, and repair its plan accordingly when tool calls fail."}, "model": "deepseek-v4-pro", "reviewedAt": "2026-09-14T00:43:53.625753Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "ParaRecover evaluates LLM-based parallel tool-use agents on error localization and recovery, testing whether agents detect execution errors, diagnose root causes, and repair plans when tool calls fail. It scores replanner output across structural, diagnostic, and evolutionary dimensions.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["exact coined title identity tied to benchmark evidence", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL"], "snippet": "We introduce ParaRecover, a process-level benchmark for evaluating error localization and recovery in multi-turn parallel tool-use agents."}, "familyId": "bmf_e307f345a2e1", "firstSeenAt": "2026-09-14", "heat": null, "id": "bm_pararecover_c43c0d4d", "indexedAt": "2026-09-14T00:39:56.575498Z", "industrySectors": [], "links": {"code": "https://github.com/gbw206/ParaRecover", "data": null, "paper": "https://arxiv.org/abs/2609.12345", "pdf": "https://arxiv.org/pdf/2609.12345", "project": null, "report": "https://arxiv.org/abs/2609.12345"}, "metrics": [], "motivation": "Existing agent benchmarks mainly evaluate final task success or tool-call correctness, providing limited insight into whether agents can reliably diagnose and recover from intermediate execution failures.", "name": "ParaRecover", "oneLine": "ParaRecover evaluates LLM-based parallel tool-use agents on error localization and recovery, testing whether agents detect execution errors, diagnose root causes, and repair plans when tool calls fail. It scores replanner output across structural, diagnostic, and evolutionary dimensions.", "paperTitle": "ParaRecover: A Process-Level Benchmark for Error Localization and Recovery in Parallel Tool-Use Agents", "primaryDomain": "General AI", "publication": {"evidence": "Accepted to EMNLP 2026 Main Conference", "evidenceLevel": "author-claim", "evidenceUrl": "https://arxiv.org/abs/2609.12345", "source": "arxiv-comments", "status": "acceptance_claimed", "venue": "EMNLP 2026 Main Conference", "verifiedAt": "2026-09-14T00:39:56.575498Z"}, "publications": [], "publishers": [{"name": "gbw206", "organizationType": "community", "role": "benchmark-publisher", "sourceUrl": "https://github.com/gbw206/ParaRecover"}], "ranking": {"30d": {"confidence": "High", "coverage": 0.85, "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": 0.0, "value": 0}}, "confidence": "High", "coverage": 0.85, "rank": 231, "score": 8}, "method": "attention-ranking-v13", "rank": 231, "score": 8}, "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": 0.0, "value": 0}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": 0.0, "value": 0}}, "confidence": "Medium", "coverage": 0.7, "rank": 574, "score": 15}, "method": "attention-ranking-v13", "rank": 574, "score": 15}}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-09-11", "source": {"authors": ["Bowen Guan", "Zhentao Yin", "Yanming Shen"], "categories": ["cs.LG", "cs.SE"], "id": "2609.12345", "title": "ParaRecover: A Process-Level Benchmark for Error Localization and Recovery in Parallel Tool-Use Agents", "type": "arxiv", "url": "https://arxiv.org/abs/2609.12345"}, "sourceUpdatedAt": "2026-09-11T02:09:33Z", "topics": ["cs.LG"], "venueAttempts": [{"decisionRaw": "Accepted to EMNLP 2026 Main Conference", "evidence": [{"level": "author-claim", "observedAt": "2026-09-14T00:39:56.575498Z", "rawValue": "Accepted to EMNLP 2026 Main Conference", "sourceType": "arxiv-comments", "sourceUrl": "https://arxiv.org/abs/2609.12345"}], "reviewStatus": "accepted", "venueName": "EMNLP 2026 Main Conference"}], "whyItMatters": "Parallel tool-use agents can fail mid-execution, and recovery ability is poorly measured by endpoint success alone. A process-level scoring contract helps compare agents on error diagnosis and plan repair."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
