{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:LNFII3KN22B6QAAGLUZGH7CCKL","short_pith_number":"pith:LNFII3KN","schema_version":"1.0","canonical_sha256":"5b4a846d4dd683e800065d3263fc4252d6b3b02f3d70c7c8a14e184a5acbe6e7","source":{"kind":"arxiv","id":"2512.13278","version":2},"attestation_state":"computed","paper":{"title":"AutoTool: Dynamic Tool Selection and Integration for Agentic Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Jiaru Zou, Jingrui He, Ke Shen, Ling Yang, Mengdi Wang, Mengting Ai, Sirui Chen, Yunzhe Qi","submitted_at":"2025-12-15T12:38:04Z","abstract_excerpt":"Agentic reinforcement learning has advanced large language models (LLMs) to reason through long chain-of-thought trajectories while interleaving external tool use. Existing approaches assume a fixed inventory of tools, which limits the adaptability of LLM agents to new or evolving toolsets. We present AutoTool, a training framework that equips LLM agents with dynamic tool-selection capabilities throughout their reasoning trajectories. AutoTool employs a dual-phase optimization pipeline: (i) SFT and RL-based trajectory stabilization for coherent reasoning, and (ii) KL-regularized Plackett-Luce "},"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":"2512.13278","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-12-15T12:38:04Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"36864be6454a24c48af85e727a2856f1d9e4f1101ff5ae65da64969ebfdb9611","abstract_canon_sha256":"6e2a4f29e174e71c0c58bb44335d452928ed31e8c1e26190229bbd53cdaf7a08"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-08T01:03:53.937604Z","signature_b64":"f23OMt8i7AFCskuXPk+vwfAA9fYC7hUGa0IZn7eOoI4y4ZCjSYJq7a6t8ixYOZNbVsEZYLPdZ5a6I8+xF/KCBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5b4a846d4dd683e800065d3263fc4252d6b3b02f3d70c7c8a14e184a5acbe6e7","last_reissued_at":"2026-06-08T01:03:53.936606Z","signature_status":"signed_v1","first_computed_at":"2026-06-08T01:03:53.936606Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"AutoTool: Dynamic Tool Selection and Integration for Agentic Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Jiaru Zou, Jingrui He, Ke Shen, Ling Yang, Mengdi Wang, Mengting Ai, Sirui Chen, Yunzhe Qi","submitted_at":"2025-12-15T12:38:04Z","abstract_excerpt":"Agentic reinforcement learning has advanced large language models (LLMs) to reason through long chain-of-thought trajectories while interleaving external tool use. Existing approaches assume a fixed inventory of tools, which limits the adaptability of LLM agents to new or evolving toolsets. We present AutoTool, a training framework that equips LLM agents with dynamic tool-selection capabilities throughout their reasoning trajectories. AutoTool employs a dual-phase optimization pipeline: (i) SFT and RL-based trajectory stabilization for coherent reasoning, and (ii) KL-regularized Plackett-Luce "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2512.13278","kind":"arxiv","version":2},"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/2512.13278/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":"2512.13278","created_at":"2026-06-08T01:03:53.936743+00:00"},{"alias_kind":"arxiv_version","alias_value":"2512.13278v2","created_at":"2026-06-08T01:03:53.936743+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.13278","created_at":"2026-06-08T01:03:53.936743+00:00"},{"alias_kind":"pith_short_12","alias_value":"LNFII3KN22B6","created_at":"2026-06-08T01:03:53.936743+00:00"},{"alias_kind":"pith_short_16","alias_value":"LNFII3KN22B6QAAG","created_at":"2026-06-08T01:03:53.936743+00:00"},{"alias_kind":"pith_short_8","alias_value":"LNFII3KN","created_at":"2026-06-08T01:03:53.936743+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":4,"sample":[{"citing_arxiv_id":"2605.18747","citing_title":"Code as Agent Harness","ref_index":232,"is_internal_anchor":true},{"citing_arxiv_id":"2601.12538","citing_title":"Agentic Reasoning for Large Language Models","ref_index":238,"is_internal_anchor":true},{"citing_arxiv_id":"2603.09002","citing_title":"Security Considerations for Multi-agent Systems","ref_index":58,"is_internal_anchor":true},{"citing_arxiv_id":"2604.08224","citing_title":"Externalization in LLM Agents: A Unified Review of Memory, Skills, Protocols and Harness Engineering","ref_index":204,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LNFII3KN22B6QAAGLUZGH7CCKL","json":"https://pith.science/pith/LNFII3KN22B6QAAGLUZGH7CCKL.json","graph_json":"https://pith.science/api/pith-number/LNFII3KN22B6QAAGLUZGH7CCKL/graph.json","events_json":"https://pith.science/api/pith-number/LNFII3KN22B6QAAGLUZGH7CCKL/events.json","paper":"https://pith.science/paper/LNFII3KN"},"agent_actions":{"view_html":"https://pith.science/pith/LNFII3KN22B6QAAGLUZGH7CCKL","download_json":"https://pith.science/pith/LNFII3KN22B6QAAGLUZGH7CCKL.json","view_paper":"https://pith.science/paper/LNFII3KN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2512.13278&json=true","fetch_graph":"https://pith.science/api/pith-number/LNFII3KN22B6QAAGLUZGH7CCKL/graph.json","fetch_events":"https://pith.science/api/pith-number/LNFII3KN22B6QAAGLUZGH7CCKL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LNFII3KN22B6QAAGLUZGH7CCKL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LNFII3KN22B6QAAGLUZGH7CCKL/action/storage_attestation","attest_author":"https://pith.science/pith/LNFII3KN22B6QAAGLUZGH7CCKL/action/author_attestation","sign_citation":"https://pith.science/pith/LNFII3KN22B6QAAGLUZGH7CCKL/action/citation_signature","submit_replication":"https://pith.science/pith/LNFII3KN22B6QAAGLUZGH7CCKL/action/replication_record"}},"created_at":"2026-06-08T01:03:53.936743+00:00","updated_at":"2026-06-08T01:03:53.936743+00:00"}