{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:CPRLJF3BRJNM44IW5PKJDYB5TU","short_pith_number":"pith:CPRLJF3B","schema_version":"1.0","canonical_sha256":"13e2b497618a5ace7116ebd491e03d9d2f9eb172f5356d58bf89b08b4a32ba0b","source":{"kind":"arxiv","id":"2606.05806","version":1},"attestation_state":"computed","paper":{"title":"When Tools Fail: Benchmarking Dynamic Replanning and Anomaly Recovery in LLM Agents","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Dawei Yin, Dongsheng Zhu, Lingyong Yan, Shuaiqiang Wang, Xiang Li, Xuchen Ma, Yucheng Shen, Yukun Zhao","submitted_at":"2026-06-04T07:38:46Z","abstract_excerpt":"Existing benchmarks evaluate Tool-Integrated Reasoning (TIR) in LLMs on idealized ''happy paths'', largely overlooking real-world tool failures. We introduce ToolMaze, a benchmark for dynamic path discovery and error recovery in TIR agents. To separate systematic replanning from blind trial-and-error, ToolMaze adopts a two-dimensional design: DAG-based topological complexity and a $2 \\times 2$ taxonomy of tool perturbations (explicit/implicit, transient/permanent). Evaluations show that perturbations degrade performance across nearly all models, with the sharpest drops under implicit semantic "},"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":"2606.05806","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-06-04T07:38:46Z","cross_cats_sorted":[],"title_canon_sha256":"00de78426a68a168cfbdd8e9ca59088bd9e5cc80225fe73104b07e4b3fab4926","abstract_canon_sha256":"bbea97d763beaa95a91459ab8d79c2c6906448df62c86f10dee409e137447c30"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-05T01:15:03.970679Z","signature_b64":"f8w+vzXRTDzjgkV21i/sR2l19bOo4pCJcFZo/fOnEHAfS/pzgx0z9thJZldS/8lrw/Vun3aeqb1MwEQxvLG0BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"13e2b497618a5ace7116ebd491e03d9d2f9eb172f5356d58bf89b08b4a32ba0b","last_reissued_at":"2026-06-05T01:15:03.970213Z","signature_status":"signed_v1","first_computed_at":"2026-06-05T01:15:03.970213Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"When Tools Fail: Benchmarking Dynamic Replanning and Anomaly Recovery in LLM Agents","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Dawei Yin, Dongsheng Zhu, Lingyong Yan, Shuaiqiang Wang, Xiang Li, Xuchen Ma, Yucheng Shen, Yukun Zhao","submitted_at":"2026-06-04T07:38:46Z","abstract_excerpt":"Existing benchmarks evaluate Tool-Integrated Reasoning (TIR) in LLMs on idealized ''happy paths'', largely overlooking real-world tool failures. We introduce ToolMaze, a benchmark for dynamic path discovery and error recovery in TIR agents. To separate systematic replanning from blind trial-and-error, ToolMaze adopts a two-dimensional design: DAG-based topological complexity and a $2 \\times 2$ taxonomy of tool perturbations (explicit/implicit, transient/permanent). Evaluations show that perturbations degrade performance across nearly all models, with the sharpest drops under implicit semantic "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.05806","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/2606.05806/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":"2606.05806","created_at":"2026-06-05T01:15:03.970277+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.05806v1","created_at":"2026-06-05T01:15:03.970277+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.05806","created_at":"2026-06-05T01:15:03.970277+00:00"},{"alias_kind":"pith_short_12","alias_value":"CPRLJF3BRJNM","created_at":"2026-06-05T01:15:03.970277+00:00"},{"alias_kind":"pith_short_16","alias_value":"CPRLJF3BRJNM44IW","created_at":"2026-06-05T01:15:03.970277+00:00"},{"alias_kind":"pith_short_8","alias_value":"CPRLJF3B","created_at":"2026-06-05T01:15:03.970277+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CPRLJF3BRJNM44IW5PKJDYB5TU","json":"https://pith.science/pith/CPRLJF3BRJNM44IW5PKJDYB5TU.json","graph_json":"https://pith.science/api/pith-number/CPRLJF3BRJNM44IW5PKJDYB5TU/graph.json","events_json":"https://pith.science/api/pith-number/CPRLJF3BRJNM44IW5PKJDYB5TU/events.json","paper":"https://pith.science/paper/CPRLJF3B"},"agent_actions":{"view_html":"https://pith.science/pith/CPRLJF3BRJNM44IW5PKJDYB5TU","download_json":"https://pith.science/pith/CPRLJF3BRJNM44IW5PKJDYB5TU.json","view_paper":"https://pith.science/paper/CPRLJF3B","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.05806&json=true","fetch_graph":"https://pith.science/api/pith-number/CPRLJF3BRJNM44IW5PKJDYB5TU/graph.json","fetch_events":"https://pith.science/api/pith-number/CPRLJF3BRJNM44IW5PKJDYB5TU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CPRLJF3BRJNM44IW5PKJDYB5TU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CPRLJF3BRJNM44IW5PKJDYB5TU/action/storage_attestation","attest_author":"https://pith.science/pith/CPRLJF3BRJNM44IW5PKJDYB5TU/action/author_attestation","sign_citation":"https://pith.science/pith/CPRLJF3BRJNM44IW5PKJDYB5TU/action/citation_signature","submit_replication":"https://pith.science/pith/CPRLJF3BRJNM44IW5PKJDYB5TU/action/replication_record"}},"created_at":"2026-06-05T01:15:03.970277+00:00","updated_at":"2026-06-05T01:15:03.970277+00:00"}