{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2605.26302", "kind": "paper", "url": "https://arxiv.org/abs/2605.26302"}, {"id": "gh:vita-group/agingbench", "kind": "repo", "url": "https://github.com/VITA-Group/AgingBench"}], "authors": ["Jianing Zhu", "Yeonju Ro", "John Robertson", "Kevin Wang", "Junbo Li", "Haris Vikalo", "Aditya Akella", "Zhangyang Wang"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2605.26302", "source_url": "https://arxiv.org/abs/2605.26302"}, "categories": ["Agents", "General AI", "Language & Knowledge"], "description": {"en": "Evaluates the reliability of deployed AI agents over extended operational lifetimes. The benchmark organizes agent aging into four mechanisms—compression, interference, revision, and maintenance—and uses temporal dependency graphs and paired counterfactual probes to produce diagnostic profiles of the memory pipeline's write, retrieval, and utilization stages. It includes multiple scenarios and memory policies, with scoring based on task performance over sessions."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2605.26302", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2605.26302", "title": "AgingBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["The University of Texas at Austin"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2605.26302", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2605.26302", "source_url": "https://openalex.org/W7162560177"}, "key": "claire-radar:2605.26302", "modality": null, "name": "AgingBench", "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": "unknown", "origin_record_id": "bm_agingbench_6fa9d4b4", "origin_source": "arxiv", "recognition_confidence": "0.85", "record_sha256": "adb9b93ff99728c889551ce26e3ffb199fe3e44084e602c90f202d3f32359dae", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2605.26302"}, "publisher": null, "released": "2026-05-25", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2605.26302", "source_url": "https://arxiv.org/abs/2605.26302"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2605-26302", "source": "claire_radar", "source_benchmark_id": "2605.26302", "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/VITA-Group/AgingBench", "githubScope": "benchmark_repo", "githubStars": 16, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 28, "hfPaperUrl": "https://huggingface.co/papers/2605.26302", "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": "2026-09-30", "state": "stale"}}, "status": "fresh"}, "capabilities": [], "confidence": "Medium", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "84f93073e07349155d3e1a6a3f72cecefabf341b07baf03a34b6f6ae578b3542", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "Evaluates the reliability of deployed AI agents over extended operational lifetimes. The benchmark organizes agent aging into four mechanisms—compression, interference, revision, and maintenance—and uses temporal dependency graphs and paired counterfactual probes to produce diagnostic profiles of the memory pipeline's write, retrieval, and utilization stages. It includes multiple scenarios and memory policies, with scoring based on task performance over sessions.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact named benchmark artifact released in abstract", "benchmark release stated in one sentence"], "snippet": "We introduce AgingBench, a longitudinal reliability benchmark for agent lifespan engineering: measuring not only whether deployed agents degrade, but what form the degradation takes and where repair should target."}, "familyId": "bmf_425e6cf5e16c", "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/VITA-Group/AgingBench", "url": "https://github.com/VITA-Group/AgingBench"}, {"basis": "existing-metric-association", "linkedUrl": "https://github.com/VITA-Group/AgingBench", "url": "https://huggingface.co/papers/2605.26302"}, {"basis": "huggingface-paper-association", "linkedUrl": "https://github.com/VITA-Group/AgingBench", "url": "https://huggingface.co/api/papers/2605.26302"}], "scope": "benchmark_repo", "url": "https://github.com/VITA-Group/AgingBench", "verification": "source-linked"}], "status": "linked"}, "heat": null, "id": "bm_agingbench_6fa9d4b4", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/VITA-Group/AgingBench", "data": null, "hfPaper": "https://huggingface.co/papers/2605.26302", "paper": "https://arxiv.org/abs/2605.26302", "pdf": "https://arxiv.org/pdf/2605.26302", "project": null, "report": "https://arxiv.org/abs/2605.26302"}, "metrics": [], "motivation": "Long-lived AI agents are increasingly deployed as persistent operational systems, yet they are still evaluated like freshly initialized models.", "name": "AgingBench", "oneLine": "Evaluates the reliability of deployed AI agents over extended operational lifetimes. The benchmark organizes agent aging into four mechanisms—compression, interference, revision, and maintenance—and uses temporal dependency graphs and paired counterfactual probes to produce diagnostic profiles of the memory pipeline's write, retrieval, and utilization stages. It includes multiple scenarios and memory policies, with scoring based on task performance over sessions.", "paperTitle": "Your Agents Are Aging Too: Agent Lifespan Engineering for Deployed Systems", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2605.26302", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "VITA Group", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://github.com/VITA-Group/AgingBench"}], "ranking": {}, "readiness": "Runnable", "recognitionConfidence": 0.85, "relation": "introduces", "releasedAt": "2026-05-25", "source": {"authors": ["Jianing Zhu", "Yeonju Ro", "John Robertson", "Kevin Wang", "Junbo Li", "Haris Vikalo", "Aditya Akella", "Zhangyang Wang"], "categories": ["cs.AI", "cs.CL", "cs.MA"], "id": "2605.26302", "title": "Your Agents Are Aging Too: Agent Lifespan Engineering for Deployed Systems", "type": "arxiv", "url": "https://arxiv.org/abs/2605.26302"}, "sourceUpdatedAt": "2026-05-25T19:55:12Z", "topics": ["Agents"], "venueAttempts": [], "whyItMatters": "Existing agent evaluations are snapshot-based and ignore how agents degrade after deployment. This benchmark measures longevity and provides mechanism-level diagnosis, enabling stage-targeted repair and more dependable agent deployment."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
