{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:UOMA4LXD6VIRP64FHZSBKB7MID","short_pith_number":"pith:UOMA4LXD","schema_version":"1.0","canonical_sha256":"a3980e2ee3f55117fb853e641507ec40e100e7f739b4aefd92d6790d076a4a5a","source":{"kind":"arxiv","id":"2504.15257","version":1},"attestation_state":"computed","paper":{"title":"FlowReasoner: Reinforcing Query-Level Meta-Agents","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Bryan Hooi, Chao Du, Hongcheng Gao, Longxu Dou, Min Lin, Tianyu Pang, Yue Liu, Yufei He, Zhijie Deng","submitted_at":"2025-04-21T17:35:42Z","abstract_excerpt":"This paper proposes a query-level meta-agent named FlowReasoner to automate the design of query-level multi-agent systems, i.e., one system per user query. Our core idea is to incentivize a reasoning-based meta-agent via external execution feedback. Concretely, by distilling DeepSeek R1, we first endow the basic reasoning ability regarding the generation of multi-agent systems to FlowReasoner. Then, we further enhance it via reinforcement learning (RL) with external execution feedback. A multi-purpose reward is designed to guide the RL training from aspects of performance, complexity, and effi"},"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":"2504.15257","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-04-21T17:35:42Z","cross_cats_sorted":[],"title_canon_sha256":"aa5da8d0e9fe98826f6fe41c1cc2df03b20fc0098594b770d088907a62680dfa","abstract_canon_sha256":"540009cc8de90762d8ab6f4c406624ef90563a0462093f17e54c7aca549c7623"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:51:58.876495Z","signature_b64":"Kd12ZuWa7fL1M5j95YHZKBsWI/4OVxJr/GO3RjJwFF2/8DBc/1hLRZ71nBfmRDBedLlAJmAR352AmEBl2njdDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a3980e2ee3f55117fb853e641507ec40e100e7f739b4aefd92d6790d076a4a5a","last_reissued_at":"2026-07-05T10:51:58.876001Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:51:58.876001Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"FlowReasoner: Reinforcing Query-Level Meta-Agents","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Bryan Hooi, Chao Du, Hongcheng Gao, Longxu Dou, Min Lin, Tianyu Pang, Yue Liu, Yufei He, Zhijie Deng","submitted_at":"2025-04-21T17:35:42Z","abstract_excerpt":"This paper proposes a query-level meta-agent named FlowReasoner to automate the design of query-level multi-agent systems, i.e., one system per user query. Our core idea is to incentivize a reasoning-based meta-agent via external execution feedback. Concretely, by distilling DeepSeek R1, we first endow the basic reasoning ability regarding the generation of multi-agent systems to FlowReasoner. Then, we further enhance it via reinforcement learning (RL) with external execution feedback. A multi-purpose reward is designed to guide the RL training from aspects of performance, complexity, and effi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.15257","kind":"arxiv","version":1},"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/2504.15257/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":"2504.15257","created_at":"2026-07-05T10:51:58.876060+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.15257v1","created_at":"2026-07-05T10:51:58.876060+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.15257","created_at":"2026-07-05T10:51:58.876060+00:00"},{"alias_kind":"pith_short_12","alias_value":"UOMA4LXD6VIR","created_at":"2026-07-05T10:51:58.876060+00:00"},{"alias_kind":"pith_short_16","alias_value":"UOMA4LXD6VIRP64F","created_at":"2026-07-05T10:51:58.876060+00:00"},{"alias_kind":"pith_short_8","alias_value":"UOMA4LXD","created_at":"2026-07-05T10:51:58.876060+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":15,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2607.07989","citing_title":"Who Broke the System? Failure Localization in LLM-Based Multi-Agent Systems","ref_index":64,"is_internal_anchor":true},{"citing_arxiv_id":"2607.06935","citing_title":"Mathematical methods of reinforcement learning","ref_index":118,"is_internal_anchor":true},{"citing_arxiv_id":"2606.23537","citing_title":"SQLConductor: Search-to-Policy Learning for Step-wise Text-to-SQL Orchestration","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2606.12191","citing_title":"Agentic Environment Engineering for Large Language Models: A Survey of Environment Modeling, Synthesis, Evaluation, and Application","ref_index":279,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08656","citing_title":"From Player to Master: Enhancing Test-Time Learning of LLM Agents via Reinforcement Learning over Memory","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2605.29790","citing_title":"Evolve as a Team: Collaborative Self-Evolution for LLM-based Multi-Agent Systems","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2605.22505","citing_title":"Towards Direct Evaluation of Harness Optimizers via Priority Ranking","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2510.05746","citing_title":"ARM: Discovering Agentic Reasoning Modules for Generalizable Multi-Agent Systems","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08401","citing_title":"AIPO: Learning to Reason from Active Interaction","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2508.07407","citing_title":"A Comprehensive Survey of Self-Evolving AI Agents: A New Paradigm Bridging Foundation Models and Lifelong Agentic Systems","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14483","citing_title":"LEMON: Learning Executable Multi-Agent Orchestration via Counterfactual Reinforcement Learning","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2603.25111","citing_title":"SEVerA: Verified Synthesis of Self-Evolving Agents","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2507.21046","citing_title":"A Survey of Self-Evolving Agents: What, When, How, and Where to Evolve on the Path to Artificial Super Intelligence","ref_index":274,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08401","citing_title":"AIPO: Learning to Reason from Active Interaction","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17503","citing_title":"SkillGraph: Self-Evolving Multi-Agent Collaboration with Multimodal Graph Topology","ref_index":11,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UOMA4LXD6VIRP64FHZSBKB7MID","json":"https://pith.science/pith/UOMA4LXD6VIRP64FHZSBKB7MID.json","graph_json":"https://pith.science/api/pith-number/UOMA4LXD6VIRP64FHZSBKB7MID/graph.json","events_json":"https://pith.science/api/pith-number/UOMA4LXD6VIRP64FHZSBKB7MID/events.json","paper":"https://pith.science/paper/UOMA4LXD"},"agent_actions":{"view_html":"https://pith.science/pith/UOMA4LXD6VIRP64FHZSBKB7MID","download_json":"https://pith.science/pith/UOMA4LXD6VIRP64FHZSBKB7MID.json","view_paper":"https://pith.science/paper/UOMA4LXD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.15257&json=true","fetch_graph":"https://pith.science/api/pith-number/UOMA4LXD6VIRP64FHZSBKB7MID/graph.json","fetch_events":"https://pith.science/api/pith-number/UOMA4LXD6VIRP64FHZSBKB7MID/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UOMA4LXD6VIRP64FHZSBKB7MID/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UOMA4LXD6VIRP64FHZSBKB7MID/action/storage_attestation","attest_author":"https://pith.science/pith/UOMA4LXD6VIRP64FHZSBKB7MID/action/author_attestation","sign_citation":"https://pith.science/pith/UOMA4LXD6VIRP64FHZSBKB7MID/action/citation_signature","submit_replication":"https://pith.science/pith/UOMA4LXD6VIRP64FHZSBKB7MID/action/replication_record"}},"created_at":"2026-07-05T10:51:58.876060+00:00","updated_at":"2026-07-05T10:51:58.876060+00:00"}