{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:BWISWDEK3T57ZKENDJSNQ5JV7C","short_pith_number":"pith:BWISWDEK","schema_version":"1.0","canonical_sha256":"0d912b0c8adcfbfca88d1a64d87535f8a83cd34f01fbeadbb115992829819ac9","source":{"kind":"arxiv","id":"2510.18586","version":3},"attestation_state":"computed","paper":{"title":"TokenCake: A KV-Cache-centric Serving Framework for LLM-based Multi-Agent Applications","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Feiyang Wu, Teng Ma, Youwei Zhuo, Zhuohang Bian, Zhuoran Li","submitted_at":"2025-10-21T12:39:32Z","abstract_excerpt":"Large Language Models (LLMs) are increasingly deployed in complex multi-agent applications that rely on external function calls. This workload creates severe performance challenges for the KV Cache: spatial contention leads to the eviction of critical agents' caches and temporal underutilization leaves the cache of agents stalled on long-running function calls idling in GPU memory. We present TokenCake, a KV-Cache-centric serving framework that bridges this gap by co-optimizing scheduling and memory management through an agent-aware design. TokenCake's Temporal Scheduler employs an event-drive"},"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":"2510.18586","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.DC","submitted_at":"2025-10-21T12:39:32Z","cross_cats_sorted":[],"title_canon_sha256":"3dca99c553a285381324dd89f2405959c78ad8c4e95a0402ffc9f6fd3bb9be59","abstract_canon_sha256":"c9a566109f70337958dd27976f81d7edbdb322bef154f4ad1141297e13168cda"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-21T01:05:11.240468Z","signature_b64":"T7EHO5XcKqAmrP/TTOISZqXLKTtfDmpNoSh/ossm7eogXO6YF3yHEb1ZkAA+BchalnYnAUrplRS3z27iw+IFDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0d912b0c8adcfbfca88d1a64d87535f8a83cd34f01fbeadbb115992829819ac9","last_reissued_at":"2026-05-21T01:05:11.239509Z","signature_status":"signed_v1","first_computed_at":"2026-05-21T01:05:11.239509Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TokenCake: A KV-Cache-centric Serving Framework for LLM-based Multi-Agent Applications","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Feiyang Wu, Teng Ma, Youwei Zhuo, Zhuohang Bian, Zhuoran Li","submitted_at":"2025-10-21T12:39:32Z","abstract_excerpt":"Large Language Models (LLMs) are increasingly deployed in complex multi-agent applications that rely on external function calls. This workload creates severe performance challenges for the KV Cache: spatial contention leads to the eviction of critical agents' caches and temporal underutilization leaves the cache of agents stalled on long-running function calls idling in GPU memory. We present TokenCake, a KV-Cache-centric serving framework that bridges this gap by co-optimizing scheduling and memory management through an agent-aware design. TokenCake's Temporal Scheduler employs an event-drive"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2510.18586","kind":"arxiv","version":3},"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/2510.18586/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":"2510.18586","created_at":"2026-05-21T01:05:11.239645+00:00"},{"alias_kind":"arxiv_version","alias_value":"2510.18586v3","created_at":"2026-05-21T01:05:11.239645+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2510.18586","created_at":"2026-05-21T01:05:11.239645+00:00"},{"alias_kind":"pith_short_12","alias_value":"BWISWDEK3T57","created_at":"2026-05-21T01:05:11.239645+00:00"},{"alias_kind":"pith_short_16","alias_value":"BWISWDEK3T57ZKEN","created_at":"2026-05-21T01:05:11.239645+00:00"},{"alias_kind":"pith_short_8","alias_value":"BWISWDEK","created_at":"2026-05-21T01:05:11.239645+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":4,"sample":[{"citing_arxiv_id":"2604.03143","citing_title":"TokenDance: Scaling Multi-Agent LLM Serving via Collective KV Cache Sharing","ref_index":2,"is_internal_anchor":true},{"citing_arxiv_id":"2604.06370","citing_title":"ForkKV: Scaling Multi-LoRA Agent Serving via Copy-on-Write Disaggregated KV Cache","ref_index":6,"is_internal_anchor":true},{"citing_arxiv_id":"2604.15186","citing_title":"Scepsy: Serving Agentic Workflows Using Aggregate LLM Pipelines","ref_index":3,"is_internal_anchor":true},{"citing_arxiv_id":"2604.17353","citing_title":"Hive: A Multi-Agent Infrastructure for Algorithm- and Task-Level Scaling","ref_index":5,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BWISWDEK3T57ZKENDJSNQ5JV7C","json":"https://pith.science/pith/BWISWDEK3T57ZKENDJSNQ5JV7C.json","graph_json":"https://pith.science/api/pith-number/BWISWDEK3T57ZKENDJSNQ5JV7C/graph.json","events_json":"https://pith.science/api/pith-number/BWISWDEK3T57ZKENDJSNQ5JV7C/events.json","paper":"https://pith.science/paper/BWISWDEK"},"agent_actions":{"view_html":"https://pith.science/pith/BWISWDEK3T57ZKENDJSNQ5JV7C","download_json":"https://pith.science/pith/BWISWDEK3T57ZKENDJSNQ5JV7C.json","view_paper":"https://pith.science/paper/BWISWDEK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2510.18586&json=true","fetch_graph":"https://pith.science/api/pith-number/BWISWDEK3T57ZKENDJSNQ5JV7C/graph.json","fetch_events":"https://pith.science/api/pith-number/BWISWDEK3T57ZKENDJSNQ5JV7C/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BWISWDEK3T57ZKENDJSNQ5JV7C/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BWISWDEK3T57ZKENDJSNQ5JV7C/action/storage_attestation","attest_author":"https://pith.science/pith/BWISWDEK3T57ZKENDJSNQ5JV7C/action/author_attestation","sign_citation":"https://pith.science/pith/BWISWDEK3T57ZKENDJSNQ5JV7C/action/citation_signature","submit_replication":"https://pith.science/pith/BWISWDEK3T57ZKENDJSNQ5JV7C/action/replication_record"}},"created_at":"2026-05-21T01:05:11.239645+00:00","updated_at":"2026-05-21T01:05:11.239645+00:00"}