{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:HRPFSIX2DSGAC5MMRNT32Q7RO2","short_pith_number":"pith:HRPFSIX2","canonical_record":{"source":{"id":"2605.09341","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.MA","submitted_at":"2026-05-10T05:43:12Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"8ab520ecd26b987d45e00875fc65cc7710a0c31fa2815d1bb657a5df8888ec4a","abstract_canon_sha256":"85d331b206d5881ef6b3f765191bccd938f96773d5ecf53abf08b29499e2a0a7"},"schema_version":"1.0"},"canonical_sha256":"3c5e5922fa1c8c01758c8b67bd43f176b8de161a9609aa780511427e9192eb0b","source":{"kind":"arxiv","id":"2605.09341","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.09341","created_at":"2026-05-20T00:03:16Z"},{"alias_kind":"arxiv_version","alias_value":"2605.09341v2","created_at":"2026-05-20T00:03:16Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.09341","created_at":"2026-05-20T00:03:16Z"},{"alias_kind":"pith_short_12","alias_value":"HRPFSIX2DSGA","created_at":"2026-05-20T00:03:16Z"},{"alias_kind":"pith_short_16","alias_value":"HRPFSIX2DSGAC5MM","created_at":"2026-05-20T00:03:16Z"},{"alias_kind":"pith_short_8","alias_value":"HRPFSIX2","created_at":"2026-05-20T00:03:16Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:HRPFSIX2DSGAC5MMRNT32Q7RO2","target":"record","payload":{"canonical_record":{"source":{"id":"2605.09341","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.MA","submitted_at":"2026-05-10T05:43:12Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"8ab520ecd26b987d45e00875fc65cc7710a0c31fa2815d1bb657a5df8888ec4a","abstract_canon_sha256":"85d331b206d5881ef6b3f765191bccd938f96773d5ecf53abf08b29499e2a0a7"},"schema_version":"1.0"},"canonical_sha256":"3c5e5922fa1c8c01758c8b67bd43f176b8de161a9609aa780511427e9192eb0b","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:03:16.336328Z","signature_b64":"+UqAdWHopSQB8vkGbwu6GL9sNL743xBRZt5wqdK9xdFR1F/WGfzRKRAUhLAsp2OwbIEhwh7eXF2Vq1CVpk0HAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3c5e5922fa1c8c01758c8b67bd43f176b8de161a9609aa780511427e9192eb0b","last_reissued_at":"2026-05-20T00:03:16.335457Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:03:16.335457Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.09341","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:03:16Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"SQuHtAXa70QuwsuOlY/pDhTWyuRAwdg8BSbQeIDtJYXFIcohSPhXcwXzdkreMz2FIcIgo/DL8f+9uF1VPlEaAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-20T12:02:26.914764Z"},"content_sha256":"52b2345a25c76bd07f22a61890e9cb64b44cf514682744483ca3c40c1dcc686b","schema_version":"1.0","event_id":"sha256:52b2345a25c76bd07f22a61890e9cb64b44cf514682744483ca3c40c1dcc686b"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:HRPFSIX2DSGAC5MMRNT32Q7RO2","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"SkillMAS: Skill Co-Evolution with LLM-based Multi-Agent System","license":"http://creativecommons.org/licenses/by/4.0/","headline":"SkillMAS couples skill evolution with multi-agent system restructuring through utility learning and evidence gating.","cross_cats":["cs.CL"],"primary_cat":"cs.MA","authors_text":"Jianghao Lin, Jiaye Gao, Jun Wang, Shuai Pan, Te Gao, Weinan Zhang, Weiwen Liu, Yixiang Liu, Yong Yu, Zhihui Fu","submitted_at":"2026-05-10T05:43:12Z","abstract_excerpt":"Large language model (LLM) agent systems are increasingly expected to improve after deployment, but existing work often decouples two adaptation targets: skill evolution and multi-agent system (MAS) restructuring. This separation can create organization bottlenecks, context pressure, and mis-specialization. We present SkillMAS, a non-parametric framework for adaptive specialization in multi-agent systems that couples skill evolution with MAS restructuring. SkillMAS uses Utility Learning to assign credit from verified execution traces, bounded skill evolution to refine reusable procedures witho"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"SkillMAS uses Utility Learning to assign credit from verified execution traces, bounded skill evolution to refine reusable procedures without unfiltered library growth, and evidence-gated MAS restructuring when retained failures and Executor Utility indicate a structural mismatch. Across embodied manipulation, command-line execution, and retail workflows, SkillMAS is competitive under the reported harnesses while clarifying how post-deployment specialization is attributed, updated, and applied.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That verified execution traces provide unbiased credit assignment for skill utility and that evidence from retained failures reliably signals when MAS restructuring is needed, without the framework introducing its own biases in trace selection or utility thresholds.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"SkillMAS couples skill evolution with multi-agent system restructuring in LLM agents through utility learning from traces, bounded refinement, and evidence-gated changes, achieving competitive results on manipulation, CLI, and retail tasks.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"SkillMAS couples skill evolution with multi-agent system restructuring through utility learning and evidence gating.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"01fa1ae615c78ba9be92a76303a8c4e3bd98dd26dd9e737f3f886557319ae789"},"source":{"id":"2605.09341","kind":"arxiv","version":2},"verdict":{"id":"fb12214f-ac33-494f-acc5-2bb3b5414b46","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-12T01:51:07.810212Z","strongest_claim":"SkillMAS uses Utility Learning to assign credit from verified execution traces, bounded skill evolution to refine reusable procedures without unfiltered library growth, and evidence-gated MAS restructuring when retained failures and Executor Utility indicate a structural mismatch. Across embodied manipulation, command-line execution, and retail workflows, SkillMAS is competitive under the reported harnesses while clarifying how post-deployment specialization is attributed, updated, and applied.","one_line_summary":"SkillMAS couples skill evolution with multi-agent system restructuring in LLM agents through utility learning from traces, bounded refinement, and evidence-gated changes, achieving competitive results on manipulation, CLI, and retail tasks.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That verified execution traces provide unbiased credit assignment for skill utility and that evidence from retained failures reliably signals when MAS restructuring is needed, without the framework introducing its own biases in trace selection or utility thresholds.","pith_extraction_headline":"SkillMAS couples skill evolution with multi-agent system restructuring through utility learning and evidence gating."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.09341/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-19T20:33:24.553650Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-19T13:01:18.586450Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T10:20:18.107322Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"fc15bdf4094cad65b08baf6d70db1a9e087e15f23c204d85fed439d25781291d"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"8c4661ce03dca6ab55af928bbbfddc340c96ae3d79ceee16d543e00972c4c883"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"fb12214f-ac33-494f-acc5-2bb3b5414b46"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:03:16Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"qxDvGfWdPGC9KObYKrPitm2FaLmt6bbhFzBhxqoa7Q/Se5pEsq1STnOxSJmdAJUzKqbj9ugAV58tLAYiV9nrCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-20T12:02:26.915285Z"},"content_sha256":"9642149ad9da2b54e83a3f97b2322089e28137b4ff5c12e2380c7cacdda1c8ee","schema_version":"1.0","event_id":"sha256:9642149ad9da2b54e83a3f97b2322089e28137b4ff5c12e2380c7cacdda1c8ee"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/HRPFSIX2DSGAC5MMRNT32Q7RO2/bundle.json","state_url":"https://pith.science/pith/HRPFSIX2DSGAC5MMRNT32Q7RO2/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/HRPFSIX2DSGAC5MMRNT32Q7RO2/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-05-20T12:02:26Z","links":{"resolver":"https://pith.science/pith/HRPFSIX2DSGAC5MMRNT32Q7RO2","bundle":"https://pith.science/pith/HRPFSIX2DSGAC5MMRNT32Q7RO2/bundle.json","state":"https://pith.science/pith/HRPFSIX2DSGAC5MMRNT32Q7RO2/state.json","well_known_bundle":"https://pith.science/.well-known/pith/HRPFSIX2DSGAC5MMRNT32Q7RO2/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:HRPFSIX2DSGAC5MMRNT32Q7RO2","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"85d331b206d5881ef6b3f765191bccd938f96773d5ecf53abf08b29499e2a0a7","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.MA","submitted_at":"2026-05-10T05:43:12Z","title_canon_sha256":"8ab520ecd26b987d45e00875fc65cc7710a0c31fa2815d1bb657a5df8888ec4a"},"schema_version":"1.0","source":{"id":"2605.09341","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.09341","created_at":"2026-05-20T00:03:16Z"},{"alias_kind":"arxiv_version","alias_value":"2605.09341v2","created_at":"2026-05-20T00:03:16Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.09341","created_at":"2026-05-20T00:03:16Z"},{"alias_kind":"pith_short_12","alias_value":"HRPFSIX2DSGA","created_at":"2026-05-20T00:03:16Z"},{"alias_kind":"pith_short_16","alias_value":"HRPFSIX2DSGAC5MM","created_at":"2026-05-20T00:03:16Z"},{"alias_kind":"pith_short_8","alias_value":"HRPFSIX2","created_at":"2026-05-20T00:03:16Z"}],"graph_snapshots":[{"event_id":"sha256:9642149ad9da2b54e83a3f97b2322089e28137b4ff5c12e2380c7cacdda1c8ee","target":"graph","created_at":"2026-05-20T00:03:16Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"SkillMAS uses Utility Learning to assign credit from verified execution traces, bounded skill evolution to refine reusable procedures without unfiltered library growth, and evidence-gated MAS restructuring when retained failures and Executor Utility indicate a structural mismatch. Across embodied manipulation, command-line execution, and retail workflows, SkillMAS is competitive under the reported harnesses while clarifying how post-deployment specialization is attributed, updated, and applied."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That verified execution traces provide unbiased credit assignment for skill utility and that evidence from retained failures reliably signals when MAS restructuring is needed, without the framework introducing its own biases in trace selection or utility thresholds."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"SkillMAS couples skill evolution with multi-agent system restructuring in LLM agents through utility learning from traces, bounded refinement, and evidence-gated changes, achieving competitive results on manipulation, CLI, and retail tasks."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"SkillMAS couples skill evolution with multi-agent system restructuring through utility learning and evidence gating."}],"snapshot_sha256":"01fa1ae615c78ba9be92a76303a8c4e3bd98dd26dd9e737f3f886557319ae789"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"8c4661ce03dca6ab55af928bbbfddc340c96ae3d79ceee16d543e00972c4c883"},"integrity":{"available":true,"clean":true,"detectors_run":[{"findings_count":0,"name":"ai_meta_artifact","ran_at":"2026-05-19T20:33:24.553650Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_title_agreement","ran_at":"2026-05-19T13:01:18.586450Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_compliance","ran_at":"2026-05-19T10:20:18.107322Z","status":"completed","version":"1.0.0"}],"endpoint":"/pith/2605.09341/integrity.json","findings":[],"snapshot_sha256":"fc15bdf4094cad65b08baf6d70db1a9e087e15f23c204d85fed439d25781291d","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language model (LLM) agent systems are increasingly expected to improve after deployment, but existing work often decouples two adaptation targets: skill evolution and multi-agent system (MAS) restructuring. This separation can create organization bottlenecks, context pressure, and mis-specialization. We present SkillMAS, a non-parametric framework for adaptive specialization in multi-agent systems that couples skill evolution with MAS restructuring. SkillMAS uses Utility Learning to assign credit from verified execution traces, bounded skill evolution to refine reusable procedures witho","authors_text":"Jianghao Lin, Jiaye Gao, Jun Wang, Shuai Pan, Te Gao, Weinan Zhang, Weiwen Liu, Yixiang Liu, Yong Yu, Zhihui Fu","cross_cats":["cs.CL"],"headline":"SkillMAS couples skill evolution with multi-agent system restructuring through utility learning and evidence gating.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.MA","submitted_at":"2026-05-10T05:43:12Z","title":"SkillMAS: Skill Co-Evolution with LLM-based Multi-Agent System"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.09341","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-12T01:51:07.810212Z","id":"fb12214f-ac33-494f-acc5-2bb3b5414b46","model_set":{"reader":"grok-4.3"},"one_line_summary":"SkillMAS couples skill evolution with multi-agent system restructuring in LLM agents through utility learning from traces, bounded refinement, and evidence-gated changes, achieving competitive results on manipulation, CLI, and retail tasks.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"SkillMAS couples skill evolution with multi-agent system restructuring through utility learning and evidence gating.","strongest_claim":"SkillMAS uses Utility Learning to assign credit from verified execution traces, bounded skill evolution to refine reusable procedures without unfiltered library growth, and evidence-gated MAS restructuring when retained failures and Executor Utility indicate a structural mismatch. Across embodied manipulation, command-line execution, and retail workflows, SkillMAS is competitive under the reported harnesses while clarifying how post-deployment specialization is attributed, updated, and applied.","weakest_assumption":"That verified execution traces provide unbiased credit assignment for skill utility and that evidence from retained failures reliably signals when MAS restructuring is needed, without the framework introducing its own biases in trace selection or utility thresholds."}},"verdict_id":"fb12214f-ac33-494f-acc5-2bb3b5414b46"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:52b2345a25c76bd07f22a61890e9cb64b44cf514682744483ca3c40c1dcc686b","target":"record","created_at":"2026-05-20T00:03:16Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"85d331b206d5881ef6b3f765191bccd938f96773d5ecf53abf08b29499e2a0a7","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.MA","submitted_at":"2026-05-10T05:43:12Z","title_canon_sha256":"8ab520ecd26b987d45e00875fc65cc7710a0c31fa2815d1bb657a5df8888ec4a"},"schema_version":"1.0","source":{"id":"2605.09341","kind":"arxiv","version":2}},"canonical_sha256":"3c5e5922fa1c8c01758c8b67bd43f176b8de161a9609aa780511427e9192eb0b","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"3c5e5922fa1c8c01758c8b67bd43f176b8de161a9609aa780511427e9192eb0b","first_computed_at":"2026-05-20T00:03:16.335457Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-20T00:03:16.335457Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"+UqAdWHopSQB8vkGbwu6GL9sNL743xBRZt5wqdK9xdFR1F/WGfzRKRAUhLAsp2OwbIEhwh7eXF2Vq1CVpk0HAA==","signature_status":"signed_v1","signed_at":"2026-05-20T00:03:16.336328Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.09341","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:52b2345a25c76bd07f22a61890e9cb64b44cf514682744483ca3c40c1dcc686b","sha256:9642149ad9da2b54e83a3f97b2322089e28137b4ff5c12e2380c7cacdda1c8ee"],"state_sha256":"a3ba0d2d1817cf171873fc4c1c338944244ca8d59c06462df6ae0b5ef4cf55ab"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"d6ZRISAiVt4xLaW1PSKnRVwI2eQ8jfPbac0hA64ghY3ZAkEN0qEHePfRjF30JDGjO8ocgHdmhaMI/B6KMdEDDg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-20T12:02:26.917907Z","bundle_sha256":"3ef5180aae01646ad34c8b8fa8e598636c45a0dc409ce545c34bbe856564d080"}}