{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:5WZFZOGZO4QZ5UFMD37ODREID4","short_pith_number":"pith:5WZFZOGZ","schema_version":"1.0","canonical_sha256":"edb25cb8d977219ed0ac1efee1c4881f37286372d9046db54f2f39dbb076a0aa","source":{"kind":"arxiv","id":"2603.28716","version":2},"attestation_state":"computed","paper":{"title":"Dynamic Dual-Granularity Skill Bank for Agentic RL","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Chengdong Xu, Dongbin Zhao, Dong Li, Linjing Li, Qichao Zhang, Songjun Tu, Xiangyuan Lan, Yaocheng Zhang","submitted_at":"2026-03-30T17:32:11Z","abstract_excerpt":"Agentic RL can benefit substantially from reusable experience, yet existing skill-based methods mainly extract trajectory-level guidance and often lack principled mechanisms for maintaining an evolving skill memory. We propose D2Skill, a dynamic dual-granularity skill bank for agentic RL that organizes reusable experience into task skills for high-level guidance and step skills for fine-grained decision support and error correction. D2Skill jointly trains the policy and skill bank through paired baseline and skill-injected rollouts under the same policy, using their performance gap to derive h"},"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":"2603.28716","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.AI","submitted_at":"2026-03-30T17:32:11Z","cross_cats_sorted":[],"title_canon_sha256":"e994a5f047903516e42dc834626d0a6fd5d9984a98fe66d21f61b0be6dc43248","abstract_canon_sha256":"488288796f9f3692b45dd334336787c5dc6471c0022a7dfaa647cc632e889851"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T02:04:09.670095Z","signature_b64":"sd9Lu/gmZoecXt1Idgx8zYdgz+ghnjzRyX6IFlJ89jM3bxSEKldvNH5Z1KNZSmn9ABMqa7A2x+lyx/W+kn/DAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"edb25cb8d977219ed0ac1efee1c4881f37286372d9046db54f2f39dbb076a0aa","last_reissued_at":"2026-05-26T02:04:09.669310Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T02:04:09.669310Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Dynamic Dual-Granularity Skill Bank for Agentic RL","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Chengdong Xu, Dongbin Zhao, Dong Li, Linjing Li, Qichao Zhang, Songjun Tu, Xiangyuan Lan, Yaocheng Zhang","submitted_at":"2026-03-30T17:32:11Z","abstract_excerpt":"Agentic RL can benefit substantially from reusable experience, yet existing skill-based methods mainly extract trajectory-level guidance and often lack principled mechanisms for maintaining an evolving skill memory. We propose D2Skill, a dynamic dual-granularity skill bank for agentic RL that organizes reusable experience into task skills for high-level guidance and step skills for fine-grained decision support and error correction. D2Skill jointly trains the policy and skill bank through paired baseline and skill-injected rollouts under the same policy, using their performance gap to derive h"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2603.28716","kind":"arxiv","version":2},"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/2603.28716/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":"2603.28716","created_at":"2026-05-26T02:04:09.669446+00:00"},{"alias_kind":"arxiv_version","alias_value":"2603.28716v2","created_at":"2026-05-26T02:04:09.669446+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.28716","created_at":"2026-05-26T02:04:09.669446+00:00"},{"alias_kind":"pith_short_12","alias_value":"5WZFZOGZO4QZ","created_at":"2026-05-26T02:04:09.669446+00:00"},{"alias_kind":"pith_short_16","alias_value":"5WZFZOGZO4QZ5UFM","created_at":"2026-05-26T02:04:09.669446+00:00"},{"alias_kind":"pith_short_8","alias_value":"5WZFZOGZ","created_at":"2026-05-26T02:04:09.669446+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2605.17790","citing_title":"STRIDE: A Self-Reflective Agent Framework for Reliable Automatic Equation Discovery","ref_index":41,"is_internal_anchor":true},{"citing_arxiv_id":"2604.14054","citing_title":"$\\pi$-Play: Multi-Agent Self-Play via Privileged Self-Distillation without External Data","ref_index":33,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5WZFZOGZO4QZ5UFMD37ODREID4","json":"https://pith.science/pith/5WZFZOGZO4QZ5UFMD37ODREID4.json","graph_json":"https://pith.science/api/pith-number/5WZFZOGZO4QZ5UFMD37ODREID4/graph.json","events_json":"https://pith.science/api/pith-number/5WZFZOGZO4QZ5UFMD37ODREID4/events.json","paper":"https://pith.science/paper/5WZFZOGZ"},"agent_actions":{"view_html":"https://pith.science/pith/5WZFZOGZO4QZ5UFMD37ODREID4","download_json":"https://pith.science/pith/5WZFZOGZO4QZ5UFMD37ODREID4.json","view_paper":"https://pith.science/paper/5WZFZOGZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2603.28716&json=true","fetch_graph":"https://pith.science/api/pith-number/5WZFZOGZO4QZ5UFMD37ODREID4/graph.json","fetch_events":"https://pith.science/api/pith-number/5WZFZOGZO4QZ5UFMD37ODREID4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5WZFZOGZO4QZ5UFMD37ODREID4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5WZFZOGZO4QZ5UFMD37ODREID4/action/storage_attestation","attest_author":"https://pith.science/pith/5WZFZOGZO4QZ5UFMD37ODREID4/action/author_attestation","sign_citation":"https://pith.science/pith/5WZFZOGZO4QZ5UFMD37ODREID4/action/citation_signature","submit_replication":"https://pith.science/pith/5WZFZOGZO4QZ5UFMD37ODREID4/action/replication_record"}},"created_at":"2026-05-26T02:04:09.669446+00:00","updated_at":"2026-05-26T02:04:09.669446+00:00"}