{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:DM6LXZ552J4LVENK4XZ3IUT3PX","short_pith_number":"pith:DM6LXZ55","canonical_record":{"source":{"id":"2605.14304","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T03:12:29Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"9ddb57cf15e2a03801f6e7d72fc85a4e06a4675eca41fe42b7bf00f536a27170","abstract_canon_sha256":"f9eee20125e076820ee708c6ede9be132a01769f7dc8b5e52a5020269aa707b8"},"schema_version":"1.0"},"canonical_sha256":"1b3cbbe7bdd278ba91aae5f3b4527b7dc709cbc0b7f29ee7b798669a79dc29be","source":{"kind":"arxiv","id":"2605.14304","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.14304","created_at":"2026-05-17T23:39:10Z"},{"alias_kind":"arxiv_version","alias_value":"2605.14304v1","created_at":"2026-05-17T23:39:10Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.14304","created_at":"2026-05-17T23:39:10Z"},{"alias_kind":"pith_short_12","alias_value":"DM6LXZ552J4L","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"DM6LXZ552J4LVENK","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"DM6LXZ55","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:DM6LXZ552J4LVENK4XZ3IUT3PX","target":"record","payload":{"canonical_record":{"source":{"id":"2605.14304","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T03:12:29Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"9ddb57cf15e2a03801f6e7d72fc85a4e06a4675eca41fe42b7bf00f536a27170","abstract_canon_sha256":"f9eee20125e076820ee708c6ede9be132a01769f7dc8b5e52a5020269aa707b8"},"schema_version":"1.0"},"canonical_sha256":"1b3cbbe7bdd278ba91aae5f3b4527b7dc709cbc0b7f29ee7b798669a79dc29be","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:39:10.065095Z","signature_b64":"uqnCqsjfRGq3RmfHrTjVYRL++LNvSwnAgVeJY9F1vEFoU5CgDeZ7R0ck12/RulUHuICF2qEWx86rS+KJoVVaDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1b3cbbe7bdd278ba91aae5f3b4527b7dc709cbc0b7f29ee7b798669a79dc29be","last_reissued_at":"2026-05-17T23:39:10.064483Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:39:10.064483Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.14304","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:39:10Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Ksd/BDaAF4KQWQxgGUnil6nB146GpNqU1CkY9FnBF2hsqcb1O+5vpovnZchLnzLvQG6AclROG3k9qubQ4CgkDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-02T20:27:58.529292Z"},"content_sha256":"6b99261f4597c893e0195fd6f7b03dd9ad46e9e028b59ba305d4562cfa762cff","schema_version":"1.0","event_id":"sha256:6b99261f4597c893e0195fd6f7b03dd9ad46e9e028b59ba305d4562cfa762cff"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:DM6LXZ552J4LVENK4XZ3IUT3PX","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Matrix-Space Reinforcement Learning for Reusing Local Transition Geometry","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Positive semidefinite matrix descriptors of trajectory segments let reinforcement learning agents reuse local transition geometry across tasks.","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Carlee Joe-Wong, Tian Lan, Zuyuan Zhang","submitted_at":"2026-05-14T03:12:29Z","abstract_excerpt":"Compositional generalization in sequential decision-making requires identifying which parts of prior rollouts remain useful for new tasks. Existing methods reuse skills or predictive models, but often overlook rich local transition geometry and dynamics. We propose Matrix-Space Reinforcement Learning (MSRL), a geometric abstraction that represents trajectory segments through positive semidefinite matrix descriptors aggregating first- and second-order statistics of lifted one-step transitions. These descriptors expose shared hidden structure, support algebraic composition in an abstract matrix "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We prove that the descriptor is well defined up to coordinate gauge, complete for the induced low-order additive signal class, additive under valid segment composition, and minimally sufficient among admissible additive descriptors. We further show that conditioning value functions on the trajectory-segment matrix yields a first-order smooth approximation of action values, enabling source-learned matrix-to-value mappings to bootstrap learning in new tasks. Empirically, MSRL achieves the best average finite-budget target AUC of 0.73.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the positive semidefinite matrix descriptors aggregating first- and second-order statistics of lifted one-step transitions actually expose shared hidden structure that supports valid algebraic composition and useful transfer across tasks.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"MSRL represents trajectory segments as PSD matrices to prove additive composition properties and bootstrap value functions for better transfer, reaching 0.73 AUC versus 0.57-0.65 baselines.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Positive semidefinite matrix descriptors of trajectory segments let reinforcement learning agents reuse local transition geometry across tasks.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"bbbc68a7184f7c3368f82f58df4aefa7bd29b6452e6b669898fc84d74ee6d349"},"source":{"id":"2605.14304","kind":"arxiv","version":1},"verdict":{"id":"198d06d8-8cda-4426-98be-68fb5b0645c0","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T01:59:09.344852Z","strongest_claim":"We prove that the descriptor is well defined up to coordinate gauge, complete for the induced low-order additive signal class, additive under valid segment composition, and minimally sufficient among admissible additive descriptors. We further show that conditioning value functions on the trajectory-segment matrix yields a first-order smooth approximation of action values, enabling source-learned matrix-to-value mappings to bootstrap learning in new tasks. Empirically, MSRL achieves the best average finite-budget target AUC of 0.73.","one_line_summary":"MSRL represents trajectory segments as PSD matrices to prove additive composition properties and bootstrap value functions for better transfer, reaching 0.73 AUC versus 0.57-0.65 baselines.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the positive semidefinite matrix descriptors aggregating first- and second-order statistics of lifted one-step transitions actually expose shared hidden structure that supports valid algebraic composition and useful transfer across tasks.","pith_extraction_headline":"Positive semidefinite matrix descriptors of trajectory segments let reinforcement learning agents reuse local transition geometry across tasks."},"references":{"count":18,"sample":[{"doi":"","year":null,"title":"Proximal Policy Optimization Algorithms","work_id":"240c67fe-d14d-4520-91c1-38a4e272ca19","ref_index":1,"cited_arxiv_id":"1707.06347","is_internal_anchor":true},{"doi":"","year":null,"title":"Cochain perspectives on temporal-difference signals for learning beyond markov dynamics.arXiv preprint arXiv:2602.06939, 2026a","work_id":"5bc83485-6f67-42fc-b5e2-ee0be098b48e","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Progressive Neural Networks","work_id":"0700d73f-b94d-4cd3-be40-086e4c4544c4","ref_index":3,"cited_arxiv_id":"1606.04671","is_internal_anchor":true},{"doi":"","year":null,"title":"Operator-Guided Invariance Learning for Continuous Reinforcement Learning","work_id":"7fa594cd-3879-4a8c-ba56-77452b6fa9d4","ref_index":4,"cited_arxiv_id":"2605.06500","is_internal_anchor":true},{"doi":"","year":null,"title":"Eigenoption Discovery through the Deep Successor Representation","work_id":"7354d0df-cbfb-4430-9ce3-8e0a96bca1df","ref_index":5,"cited_arxiv_id":"1710.11089","is_internal_anchor":true}],"resolved_work":18,"snapshot_sha256":"a486d8bb7db7210a30c236f7123cf46a60bae3f6d7dad766ce14981974d36152","internal_anchors":10},"formal_canon":{"evidence_count":3,"snapshot_sha256":"38805968c39760ee2a34a31bb5b0e8accb8bd3b3a82696975475cbe837ae0149"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"198d06d8-8cda-4426-98be-68fb5b0645c0"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:39:10Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"5gpHPc+XlTymVg6qZZO+2u0euST/kt0WeA9G9gClZ4Z4Fphlk4dI4p6t3uXEAnvC7SAE8rdugoS72YSE8P9oBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-02T20:27:58.529838Z"},"content_sha256":"2df7e004d8d12badfe8ea7c6d65fbdbbafc4f660527512521b557e4c732420f8","schema_version":"1.0","event_id":"sha256:2df7e004d8d12badfe8ea7c6d65fbdbbafc4f660527512521b557e4c732420f8"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/DM6LXZ552J4LVENK4XZ3IUT3PX/bundle.json","state_url":"https://pith.science/pith/DM6LXZ552J4LVENK4XZ3IUT3PX/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/DM6LXZ552J4LVENK4XZ3IUT3PX/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":"2026-06-02T20:27:58Z","links":{"resolver":"https://pith.science/pith/DM6LXZ552J4LVENK4XZ3IUT3PX","bundle":"https://pith.science/pith/DM6LXZ552J4LVENK4XZ3IUT3PX/bundle.json","state":"https://pith.science/pith/DM6LXZ552J4LVENK4XZ3IUT3PX/state.json","well_known_bundle":"https://pith.science/.well-known/pith/DM6LXZ552J4LVENK4XZ3IUT3PX/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:DM6LXZ552J4LVENK4XZ3IUT3PX","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":"f9eee20125e076820ee708c6ede9be132a01769f7dc8b5e52a5020269aa707b8","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T03:12:29Z","title_canon_sha256":"9ddb57cf15e2a03801f6e7d72fc85a4e06a4675eca41fe42b7bf00f536a27170"},"schema_version":"1.0","source":{"id":"2605.14304","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.14304","created_at":"2026-05-17T23:39:10Z"},{"alias_kind":"arxiv_version","alias_value":"2605.14304v1","created_at":"2026-05-17T23:39:10Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.14304","created_at":"2026-05-17T23:39:10Z"},{"alias_kind":"pith_short_12","alias_value":"DM6LXZ552J4L","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"DM6LXZ552J4LVENK","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"DM6LXZ55","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:2df7e004d8d12badfe8ea7c6d65fbdbbafc4f660527512521b557e4c732420f8","target":"graph","created_at":"2026-05-17T23:39:10Z","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":"We prove that the descriptor is well defined up to coordinate gauge, complete for the induced low-order additive signal class, additive under valid segment composition, and minimally sufficient among admissible additive descriptors. We further show that conditioning value functions on the trajectory-segment matrix yields a first-order smooth approximation of action values, enabling source-learned matrix-to-value mappings to bootstrap learning in new tasks. Empirically, MSRL achieves the best average finite-budget target AUC of 0.73."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the positive semidefinite matrix descriptors aggregating first- and second-order statistics of lifted one-step transitions actually expose shared hidden structure that supports valid algebraic composition and useful transfer across tasks."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"MSRL represents trajectory segments as PSD matrices to prove additive composition properties and bootstrap value functions for better transfer, reaching 0.73 AUC versus 0.57-0.65 baselines."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Positive semidefinite matrix descriptors of trajectory segments let reinforcement learning agents reuse local transition geometry across tasks."}],"snapshot_sha256":"bbbc68a7184f7c3368f82f58df4aefa7bd29b6452e6b669898fc84d74ee6d349"},"formal_canon":{"evidence_count":3,"snapshot_sha256":"38805968c39760ee2a34a31bb5b0e8accb8bd3b3a82696975475cbe837ae0149"},"paper":{"abstract_excerpt":"Compositional generalization in sequential decision-making requires identifying which parts of prior rollouts remain useful for new tasks. Existing methods reuse skills or predictive models, but often overlook rich local transition geometry and dynamics. We propose Matrix-Space Reinforcement Learning (MSRL), a geometric abstraction that represents trajectory segments through positive semidefinite matrix descriptors aggregating first- and second-order statistics of lifted one-step transitions. These descriptors expose shared hidden structure, support algebraic composition in an abstract matrix ","authors_text":"Carlee Joe-Wong, Tian Lan, Zuyuan Zhang","cross_cats":["cs.AI"],"headline":"Positive semidefinite matrix descriptors of trajectory segments let reinforcement learning agents reuse local transition geometry across tasks.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T03:12:29Z","title":"Matrix-Space Reinforcement Learning for Reusing Local Transition Geometry"},"references":{"count":18,"internal_anchors":10,"resolved_work":18,"sample":[{"cited_arxiv_id":"1707.06347","doi":"","is_internal_anchor":true,"ref_index":1,"title":"Proximal Policy Optimization Algorithms","work_id":"240c67fe-d14d-4520-91c1-38a4e272ca19","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Cochain perspectives on temporal-difference signals for learning beyond markov dynamics.arXiv preprint arXiv:2602.06939, 2026a","work_id":"5bc83485-6f67-42fc-b5e2-ee0be098b48e","year":null},{"cited_arxiv_id":"1606.04671","doi":"","is_internal_anchor":true,"ref_index":3,"title":"Progressive Neural Networks","work_id":"0700d73f-b94d-4cd3-be40-086e4c4544c4","year":null},{"cited_arxiv_id":"2605.06500","doi":"","is_internal_anchor":true,"ref_index":4,"title":"Operator-Guided Invariance Learning for Continuous Reinforcement Learning","work_id":"7fa594cd-3879-4a8c-ba56-77452b6fa9d4","year":null},{"cited_arxiv_id":"1710.11089","doi":"","is_internal_anchor":true,"ref_index":5,"title":"Eigenoption Discovery through the Deep Successor Representation","work_id":"7354d0df-cbfb-4430-9ce3-8e0a96bca1df","year":null}],"snapshot_sha256":"a486d8bb7db7210a30c236f7123cf46a60bae3f6d7dad766ce14981974d36152"},"source":{"id":"2605.14304","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-15T01:59:09.344852Z","id":"198d06d8-8cda-4426-98be-68fb5b0645c0","model_set":{"reader":"grok-4.3"},"one_line_summary":"MSRL represents trajectory segments as PSD matrices to prove additive composition properties and bootstrap value functions for better transfer, reaching 0.73 AUC versus 0.57-0.65 baselines.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Positive semidefinite matrix descriptors of trajectory segments let reinforcement learning agents reuse local transition geometry across tasks.","strongest_claim":"We prove that the descriptor is well defined up to coordinate gauge, complete for the induced low-order additive signal class, additive under valid segment composition, and minimally sufficient among admissible additive descriptors. We further show that conditioning value functions on the trajectory-segment matrix yields a first-order smooth approximation of action values, enabling source-learned matrix-to-value mappings to bootstrap learning in new tasks. Empirically, MSRL achieves the best average finite-budget target AUC of 0.73.","weakest_assumption":"That the positive semidefinite matrix descriptors aggregating first- and second-order statistics of lifted one-step transitions actually expose shared hidden structure that supports valid algebraic composition and useful transfer across tasks."}},"verdict_id":"198d06d8-8cda-4426-98be-68fb5b0645c0"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:6b99261f4597c893e0195fd6f7b03dd9ad46e9e028b59ba305d4562cfa762cff","target":"record","created_at":"2026-05-17T23:39:10Z","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":"f9eee20125e076820ee708c6ede9be132a01769f7dc8b5e52a5020269aa707b8","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T03:12:29Z","title_canon_sha256":"9ddb57cf15e2a03801f6e7d72fc85a4e06a4675eca41fe42b7bf00f536a27170"},"schema_version":"1.0","source":{"id":"2605.14304","kind":"arxiv","version":1}},"canonical_sha256":"1b3cbbe7bdd278ba91aae5f3b4527b7dc709cbc0b7f29ee7b798669a79dc29be","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"1b3cbbe7bdd278ba91aae5f3b4527b7dc709cbc0b7f29ee7b798669a79dc29be","first_computed_at":"2026-05-17T23:39:10.064483Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:39:10.064483Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"uqnCqsjfRGq3RmfHrTjVYRL++LNvSwnAgVeJY9F1vEFoU5CgDeZ7R0ck12/RulUHuICF2qEWx86rS+KJoVVaDQ==","signature_status":"signed_v1","signed_at":"2026-05-17T23:39:10.065095Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.14304","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:6b99261f4597c893e0195fd6f7b03dd9ad46e9e028b59ba305d4562cfa762cff","sha256:2df7e004d8d12badfe8ea7c6d65fbdbbafc4f660527512521b557e4c732420f8"],"state_sha256":"86253c83421ac13623155b9cac645906d5ebfa4e3701aa1082d425db463ebaac"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"K+vUXwKvudn4dnOeVPoINF3nj//W+8QKMkZjHm1y/WKMPEpDadyYuKlKF2H0NUafexKwzt4xzsYsDFNKjl3gDQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-02T20:27:58.532304Z","bundle_sha256":"baad4f2863c450e8b5af7cd67e43812962d3aec511bcb228d2b28636355dc410"}}