{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:MEBLKTBRUCUZJXC2ASNWFPG3LI","short_pith_number":"pith:MEBLKTBR","schema_version":"1.0","canonical_sha256":"6102b54c31a0a994dc5a049b62bcdb5a1376432f485a41b7067a71e797c55a4a","source":{"kind":"arxiv","id":"2411.13547","version":2},"attestation_state":"computed","paper":{"title":"ToolScan: A Benchmark for Characterizing Errors in Tool-Use LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.SE","authors_text":"Akshara Prabhakar, Caiming Xiong, Huan Wang, Jianguo Zhang, Juan Carlos Niebles, Juntao Tan, Liangwei Yang, Ming Zhu, Rithesh Murthy, Shelby Heinecke, Shirley Kokane, Silivo Savarese, Thai Hoang, Tian Lan, Tulika Awalgaonkar, Weiran Yao, Zhiwei Liu, Zuxin Liu","submitted_at":"2024-11-20T18:56:22Z","abstract_excerpt":"Evaluating Large Language Models (LLMs) is one of the most critical aspects of building a performant compound AI system. Since the output from LLMs propagate to downstream steps, identifying LLM errors is crucial to system performance. A common task for LLMs in AI systems is tool use. While there are several benchmark environments for evaluating LLMs on this task, they typically only give a success rate without any explanation of the failure cases. To solve this problem, we introduce TOOLSCAN, a new benchmark to identify error patterns in LLM output on tool-use tasks. Our benchmark data set co"},"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":"2411.13547","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.SE","submitted_at":"2024-11-20T18:56:22Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"ac8157c72627284b9d000983beefa5e51d5024c0e101405897782073c48ffc9c","abstract_canon_sha256":"d7c1bf8ea7ff4cb96d9ec67aa1d20dbf29a3b3b3ee942c36556fd53b2b57d952"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:27:21.876179Z","signature_b64":"2UZhLudQbXY7X172SYHUpcKIa/iZapFxFKhp8nAnJYylUlPl7aXSZNoJGUOkOHGZovP/V/toUk+mLieXauKpBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6102b54c31a0a994dc5a049b62bcdb5a1376432f485a41b7067a71e797c55a4a","last_reissued_at":"2026-07-05T11:27:21.875654Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:27:21.875654Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ToolScan: A Benchmark for Characterizing Errors in Tool-Use LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.SE","authors_text":"Akshara Prabhakar, Caiming Xiong, Huan Wang, Jianguo Zhang, Juan Carlos Niebles, Juntao Tan, Liangwei Yang, Ming Zhu, Rithesh Murthy, Shelby Heinecke, Shirley Kokane, Silivo Savarese, Thai Hoang, Tian Lan, Tulika Awalgaonkar, Weiran Yao, Zhiwei Liu, Zuxin Liu","submitted_at":"2024-11-20T18:56:22Z","abstract_excerpt":"Evaluating Large Language Models (LLMs) is one of the most critical aspects of building a performant compound AI system. Since the output from LLMs propagate to downstream steps, identifying LLM errors is crucial to system performance. A common task for LLMs in AI systems is tool use. While there are several benchmark environments for evaluating LLMs on this task, they typically only give a success rate without any explanation of the failure cases. To solve this problem, we introduce TOOLSCAN, a new benchmark to identify error patterns in LLM output on tool-use tasks. Our benchmark data set co"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.13547","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/2411.13547/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":"2411.13547","created_at":"2026-07-05T11:27:21.875727+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.13547v2","created_at":"2026-07-05T11:27:21.875727+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.13547","created_at":"2026-07-05T11:27:21.875727+00:00"},{"alias_kind":"pith_short_12","alias_value":"MEBLKTBRUCUZ","created_at":"2026-07-05T11:27:21.875727+00:00"},{"alias_kind":"pith_short_16","alias_value":"MEBLKTBRUCUZJXC2","created_at":"2026-07-05T11:27:21.875727+00:00"},{"alias_kind":"pith_short_8","alias_value":"MEBLKTBR","created_at":"2026-07-05T11:27:21.875727+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.00334","citing_title":"AgentFloor: How Far Up the tool use Ladder Can Small Open-Weight Models Go?","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2604.14723","citing_title":"Bounded Autonomy for Enterprise AI: Typed Action Contracts and Consumer-Side Execution","ref_index":20,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MEBLKTBRUCUZJXC2ASNWFPG3LI","json":"https://pith.science/pith/MEBLKTBRUCUZJXC2ASNWFPG3LI.json","graph_json":"https://pith.science/api/pith-number/MEBLKTBRUCUZJXC2ASNWFPG3LI/graph.json","events_json":"https://pith.science/api/pith-number/MEBLKTBRUCUZJXC2ASNWFPG3LI/events.json","paper":"https://pith.science/paper/MEBLKTBR"},"agent_actions":{"view_html":"https://pith.science/pith/MEBLKTBRUCUZJXC2ASNWFPG3LI","download_json":"https://pith.science/pith/MEBLKTBRUCUZJXC2ASNWFPG3LI.json","view_paper":"https://pith.science/paper/MEBLKTBR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.13547&json=true","fetch_graph":"https://pith.science/api/pith-number/MEBLKTBRUCUZJXC2ASNWFPG3LI/graph.json","fetch_events":"https://pith.science/api/pith-number/MEBLKTBRUCUZJXC2ASNWFPG3LI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MEBLKTBRUCUZJXC2ASNWFPG3LI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MEBLKTBRUCUZJXC2ASNWFPG3LI/action/storage_attestation","attest_author":"https://pith.science/pith/MEBLKTBRUCUZJXC2ASNWFPG3LI/action/author_attestation","sign_citation":"https://pith.science/pith/MEBLKTBRUCUZJXC2ASNWFPG3LI/action/citation_signature","submit_replication":"https://pith.science/pith/MEBLKTBRUCUZJXC2ASNWFPG3LI/action/replication_record"}},"created_at":"2026-07-05T11:27:21.875727+00:00","updated_at":"2026-07-05T11:27:21.875727+00:00"}