{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:NKQIMBL4OIYIAVNBGVPKHKPMRK","short_pith_number":"pith:NKQIMBL4","schema_version":"1.0","canonical_sha256":"6aa086057c72308055a1355ea3a9ec8aa3f9bbccebf304e188fa24df4f45a9d4","source":{"kind":"arxiv","id":"2606.05922","version":1},"attestation_state":"computed","paper":{"title":"Retrospective Harness Optimization: Improving LLM Agents via Self-Preference over Trajectory Rollouts","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Chin-Yew Lin, Jingying Zeng, Shujie Liu, Wenbo Pan, Xianfeng Tang, Xiangyang Zhou, Xiaohua Jia, Yan Lu","submitted_at":"2026-06-04T09:26:00Z","abstract_excerpt":"AI agents rely on a harness of skills, tools, and workflows to solve complex problems. Continually improving this harness is essential for adapting to new tasks. However, existing optimization methods typically require ground-truth validation sets, yet such labeled data is difficult to acquire in practical deployment settings. To address this problem, we introduce Retrospective Harness Optimization (RHO), a self-supervised method that optimizes the agent harness using only past trajectories. Specifically, RHO selects a diverse coreset of challenging tasks from past trajectories and re-solves t"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2606.05922","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.AI","submitted_at":"2026-06-04T09:26:00Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"67d4614003941f64ebc60159d8dfedcf9e5e73f1af758841d91fb8850cb42862","abstract_canon_sha256":"b55f873abb3aa1564e91c5f9b0e5df3385a1a6bcf97c152831af6e48a97c3d5b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-05T01:15:27.698707Z","signature_b64":"UqZvsvLbiY3ppRryRbV+SjJx41qSRJyyPMrnHTxrI+UgDsOEbpok3hO0sak9ghZNZHnxUgDe0tN4Lx5hmFjMDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6aa086057c72308055a1355ea3a9ec8aa3f9bbccebf304e188fa24df4f45a9d4","last_reissued_at":"2026-06-05T01:15:27.698311Z","signature_status":"signed_v1","first_computed_at":"2026-06-05T01:15:27.698311Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Retrospective Harness Optimization: Improving LLM Agents via Self-Preference over Trajectory Rollouts","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Chin-Yew Lin, Jingying Zeng, Shujie Liu, Wenbo Pan, Xianfeng Tang, Xiangyang Zhou, Xiaohua Jia, Yan Lu","submitted_at":"2026-06-04T09:26:00Z","abstract_excerpt":"AI agents rely on a harness of skills, tools, and workflows to solve complex problems. Continually improving this harness is essential for adapting to new tasks. However, existing optimization methods typically require ground-truth validation sets, yet such labeled data is difficult to acquire in practical deployment settings. To address this problem, we introduce Retrospective Harness Optimization (RHO), a self-supervised method that optimizes the agent harness using only past trajectories. Specifically, RHO selects a diverse coreset of challenging tasks from past trajectories and re-solves t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.05922","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2606.05922/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2606.05922","created_at":"2026-06-05T01:15:27.698377+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.05922v1","created_at":"2026-06-05T01:15:27.698377+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.05922","created_at":"2026-06-05T01:15:27.698377+00:00"},{"alias_kind":"pith_short_12","alias_value":"NKQIMBL4OIYI","created_at":"2026-06-05T01:15:27.698377+00:00"},{"alias_kind":"pith_short_16","alias_value":"NKQIMBL4OIYIAVNB","created_at":"2026-06-05T01:15:27.698377+00:00"},{"alias_kind":"pith_short_8","alias_value":"NKQIMBL4","created_at":"2026-06-05T01:15:27.698377+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NKQIMBL4OIYIAVNBGVPKHKPMRK","json":"https://pith.science/pith/NKQIMBL4OIYIAVNBGVPKHKPMRK.json","graph_json":"https://pith.science/api/pith-number/NKQIMBL4OIYIAVNBGVPKHKPMRK/graph.json","events_json":"https://pith.science/api/pith-number/NKQIMBL4OIYIAVNBGVPKHKPMRK/events.json","paper":"https://pith.science/paper/NKQIMBL4"},"agent_actions":{"view_html":"https://pith.science/pith/NKQIMBL4OIYIAVNBGVPKHKPMRK","download_json":"https://pith.science/pith/NKQIMBL4OIYIAVNBGVPKHKPMRK.json","view_paper":"https://pith.science/paper/NKQIMBL4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.05922&json=true","fetch_graph":"https://pith.science/api/pith-number/NKQIMBL4OIYIAVNBGVPKHKPMRK/graph.json","fetch_events":"https://pith.science/api/pith-number/NKQIMBL4OIYIAVNBGVPKHKPMRK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NKQIMBL4OIYIAVNBGVPKHKPMRK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NKQIMBL4OIYIAVNBGVPKHKPMRK/action/storage_attestation","attest_author":"https://pith.science/pith/NKQIMBL4OIYIAVNBGVPKHKPMRK/action/author_attestation","sign_citation":"https://pith.science/pith/NKQIMBL4OIYIAVNBGVPKHKPMRK/action/citation_signature","submit_replication":"https://pith.science/pith/NKQIMBL4OIYIAVNBGVPKHKPMRK/action/replication_record"}},"created_at":"2026-06-05T01:15:27.698377+00:00","updated_at":"2026-06-05T01:15:27.698377+00:00"}