{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:LUXMQJAZDDBLLKBMMN3NRSKRCJ","short_pith_number":"pith:LUXMQJAZ","schema_version":"1.0","canonical_sha256":"5d2ec8241918c2b5a82c6376d8c95112403e265bdc6f84b7d665610f4671b1bc","source":{"kind":"arxiv","id":"2602.13692","version":3},"attestation_state":"computed","paper":{"title":"ThunderAgent: A Simple, Fast and Program-Aware Agentic Inference System","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.MA"],"primary_cat":"cs.OS","authors_text":"Beidi Chen, Chenfeng Xu, Hao Kang, Junxiong Wang, Simran Arora, Tushar Krishna, Weili Xu, Xinyu Yang, Yinfang Chen, Ziyang Li","submitted_at":"2026-02-14T09:26:41Z","abstract_excerpt":"Large language models(LLMs) are now used to power complex multi-turn agentic workflows. Existing systems run agentic inference by loosely assembling isolated components: an LLM inference engine (e.g., vLLM) and a tool orchestrator (e.g., Kubernetes). Although agentic workflows involve multiple LLM and tool requests, these systems schedule and allocate resources separately on a per-request basis, without end-to-end knowledge of the workflow. This leads to sub-optimal management of KV cache and tool execution environments. To address the challenges, we propose ThunderAgent, a fast, simple, and p"},"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":"2602.13692","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.OS","submitted_at":"2026-02-14T09:26:41Z","cross_cats_sorted":["cs.MA"],"title_canon_sha256":"111180f07874d12b47d590b6a1c05c2eeb698715203910d2807df3f2bc613077","abstract_canon_sha256":"1b51e7d87d7fb729371c808d621053ab2a361f3a574e182e51db42f190ecf19c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-01T01:18:24.317038Z","signature_b64":"+TzmI27dfo4zmhp4fuD2EBaraoKkmUeFOUUYwM7zDleuTn71xq6GQuA7D54YyCGBeYFnZjDUL15H2nE6qXlhBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5d2ec8241918c2b5a82c6376d8c95112403e265bdc6f84b7d665610f4671b1bc","last_reissued_at":"2026-07-01T01:18:24.316528Z","signature_status":"signed_v1","first_computed_at":"2026-07-01T01:18:24.316528Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ThunderAgent: A Simple, Fast and Program-Aware Agentic Inference System","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.MA"],"primary_cat":"cs.OS","authors_text":"Beidi Chen, Chenfeng Xu, Hao Kang, Junxiong Wang, Simran Arora, Tushar Krishna, Weili Xu, Xinyu Yang, Yinfang Chen, Ziyang Li","submitted_at":"2026-02-14T09:26:41Z","abstract_excerpt":"Large language models(LLMs) are now used to power complex multi-turn agentic workflows. Existing systems run agentic inference by loosely assembling isolated components: an LLM inference engine (e.g., vLLM) and a tool orchestrator (e.g., Kubernetes). Although agentic workflows involve multiple LLM and tool requests, these systems schedule and allocate resources separately on a per-request basis, without end-to-end knowledge of the workflow. This leads to sub-optimal management of KV cache and tool execution environments. To address the challenges, we propose ThunderAgent, a fast, simple, and p"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.13692","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/2602.13692/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":"2602.13692","created_at":"2026-07-01T01:18:24.316590+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.13692v3","created_at":"2026-07-01T01:18:24.316590+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.13692","created_at":"2026-07-01T01:18:24.316590+00:00"},{"alias_kind":"pith_short_12","alias_value":"LUXMQJAZDDBL","created_at":"2026-07-01T01:18:24.316590+00:00"},{"alias_kind":"pith_short_16","alias_value":"LUXMQJAZDDBLLKBM","created_at":"2026-07-01T01:18:24.316590+00:00"},{"alias_kind":"pith_short_8","alias_value":"LUXMQJAZ","created_at":"2026-07-01T01:18:24.316590+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":8,"internal_anchor_count":8,"sample":[{"citing_arxiv_id":"2606.04315","citing_title":"Exploring Cross-Scenario Generality of Agentic Memory Systems: Diagnostics and a Strong Baseline","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2606.02964","citing_title":"Multi-Segment Attention: Enabling Efficient KV-Cache Management for Faster Large Language Model Serving","ref_index":31,"is_internal_anchor":true},{"citing_arxiv_id":"2606.00866","citing_title":"Idleness is Relative: Exploiting Tool-Call Idle Windows for Offloading in Agentic Systems with MORI","ref_index":25,"is_internal_anchor":true},{"citing_arxiv_id":"2605.26297","citing_title":"Agentic AI Workload Characteristics","ref_index":13,"is_internal_anchor":true},{"citing_arxiv_id":"2605.22502","citing_title":"Compiling Agentic Workflows into LLM Weights: Near-Frontier Quality at Two Orders of Magnitude Less Cost","ref_index":29,"is_internal_anchor":true},{"citing_arxiv_id":"2604.06296","citing_title":"AgentOpt v0.1 Technical Report: Client-Side Optimization for LLM-Based Agent","ref_index":13,"is_internal_anchor":true},{"citing_arxiv_id":"2604.16682","citing_title":"KAIROS: Stateful, Context-Aware Power-Efficient Agentic Inference Serving","ref_index":28,"is_internal_anchor":true},{"citing_arxiv_id":"2604.17353","citing_title":"Hive: A Multi-Agent Infrastructure for Algorithm- and Task-Level Scaling","ref_index":17,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LUXMQJAZDDBLLKBMMN3NRSKRCJ","json":"https://pith.science/pith/LUXMQJAZDDBLLKBMMN3NRSKRCJ.json","graph_json":"https://pith.science/api/pith-number/LUXMQJAZDDBLLKBMMN3NRSKRCJ/graph.json","events_json":"https://pith.science/api/pith-number/LUXMQJAZDDBLLKBMMN3NRSKRCJ/events.json","paper":"https://pith.science/paper/LUXMQJAZ"},"agent_actions":{"view_html":"https://pith.science/pith/LUXMQJAZDDBLLKBMMN3NRSKRCJ","download_json":"https://pith.science/pith/LUXMQJAZDDBLLKBMMN3NRSKRCJ.json","view_paper":"https://pith.science/paper/LUXMQJAZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.13692&json=true","fetch_graph":"https://pith.science/api/pith-number/LUXMQJAZDDBLLKBMMN3NRSKRCJ/graph.json","fetch_events":"https://pith.science/api/pith-number/LUXMQJAZDDBLLKBMMN3NRSKRCJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LUXMQJAZDDBLLKBMMN3NRSKRCJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LUXMQJAZDDBLLKBMMN3NRSKRCJ/action/storage_attestation","attest_author":"https://pith.science/pith/LUXMQJAZDDBLLKBMMN3NRSKRCJ/action/author_attestation","sign_citation":"https://pith.science/pith/LUXMQJAZDDBLLKBMMN3NRSKRCJ/action/citation_signature","submit_replication":"https://pith.science/pith/LUXMQJAZDDBLLKBMMN3NRSKRCJ/action/replication_record"}},"created_at":"2026-07-01T01:18:24.316590+00:00","updated_at":"2026-07-01T01:18:24.316590+00:00"}