{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:TNXM66DT4Y7CU7ZIPYEMOPUCCM","short_pith_number":"pith:TNXM66DT","schema_version":"1.0","canonical_sha256":"9b6ecf7873e63e2a7f287e08c73e821300346c157711dd8966300e6f0ea5dc88","source":{"kind":"arxiv","id":"2606.16364","version":2},"attestation_state":"computed","paper":{"title":"Looking Is Not Picking: An Attention-Segment Account of Tool-Selection Failures in LLM Agents","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CR","cs.SE"],"primary_cat":"cs.AI","authors_text":"Shiyang Chen","submitted_at":"2026-06-15T07:58:56Z","abstract_excerpt":"LLM agents mis-call tools, and the natural guess is that the model failed to see the right tool in a crowded harness. We show the opposite through a lens concurrent work sets aside -- the model's attention to labeled tool-definition segments. On real BFCL failures, by per-candidate attention argmax the model attends most to the correct tool 80% of the time (vs. 21% chance), and the gold is the under-attended segment on only 10%: it looks at the right tool and still picks wrong. This directly refutes the intuitive \"crowded-harness / lost-in-the-middle\" explanation: the failure is at the decisio"},"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":"2606.16364","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-06-15T07:58:56Z","cross_cats_sorted":["cs.CR","cs.SE"],"title_canon_sha256":"b99acedf3f69a8c18503948181b348901e530011ccf411097d5246bbb35dd810","abstract_canon_sha256":"88dcba27e04e55d0da457c31e56f1c65fc548e7bacda62e415d2b19fa19b432b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-30T01:16:31.950709Z","signature_b64":"0m+hvyu0vTgPO/8REiO4MqKs/Naf7Ttotug26n9tert9ePIJdcFUU7IcZUdU8FQ6jnGd6I9eUxVY1T8AqwYBDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9b6ecf7873e63e2a7f287e08c73e821300346c157711dd8966300e6f0ea5dc88","last_reissued_at":"2026-06-30T01:16:31.950160Z","signature_status":"signed_v1","first_computed_at":"2026-06-30T01:16:31.950160Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Looking Is Not Picking: An Attention-Segment Account of Tool-Selection Failures in LLM Agents","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CR","cs.SE"],"primary_cat":"cs.AI","authors_text":"Shiyang Chen","submitted_at":"2026-06-15T07:58:56Z","abstract_excerpt":"LLM agents mis-call tools, and the natural guess is that the model failed to see the right tool in a crowded harness. We show the opposite through a lens concurrent work sets aside -- the model's attention to labeled tool-definition segments. On real BFCL failures, by per-candidate attention argmax the model attends most to the correct tool 80% of the time (vs. 21% chance), and the gold is the under-attended segment on only 10%: it looks at the right tool and still picks wrong. This directly refutes the intuitive \"crowded-harness / lost-in-the-middle\" explanation: the failure is at the decisio"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.16364","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/2606.16364/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":"2606.16364","created_at":"2026-06-30T01:16:31.950230+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.16364v2","created_at":"2026-06-30T01:16:31.950230+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.16364","created_at":"2026-06-30T01:16:31.950230+00:00"},{"alias_kind":"pith_short_12","alias_value":"TNXM66DT4Y7C","created_at":"2026-06-30T01:16:31.950230+00:00"},{"alias_kind":"pith_short_16","alias_value":"TNXM66DT4Y7CU7ZI","created_at":"2026-06-30T01:16:31.950230+00:00"},{"alias_kind":"pith_short_8","alias_value":"TNXM66DT","created_at":"2026-06-30T01:16:31.950230+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/TNXM66DT4Y7CU7ZIPYEMOPUCCM","json":"https://pith.science/pith/TNXM66DT4Y7CU7ZIPYEMOPUCCM.json","graph_json":"https://pith.science/api/pith-number/TNXM66DT4Y7CU7ZIPYEMOPUCCM/graph.json","events_json":"https://pith.science/api/pith-number/TNXM66DT4Y7CU7ZIPYEMOPUCCM/events.json","paper":"https://pith.science/paper/TNXM66DT"},"agent_actions":{"view_html":"https://pith.science/pith/TNXM66DT4Y7CU7ZIPYEMOPUCCM","download_json":"https://pith.science/pith/TNXM66DT4Y7CU7ZIPYEMOPUCCM.json","view_paper":"https://pith.science/paper/TNXM66DT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.16364&json=true","fetch_graph":"https://pith.science/api/pith-number/TNXM66DT4Y7CU7ZIPYEMOPUCCM/graph.json","fetch_events":"https://pith.science/api/pith-number/TNXM66DT4Y7CU7ZIPYEMOPUCCM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TNXM66DT4Y7CU7ZIPYEMOPUCCM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TNXM66DT4Y7CU7ZIPYEMOPUCCM/action/storage_attestation","attest_author":"https://pith.science/pith/TNXM66DT4Y7CU7ZIPYEMOPUCCM/action/author_attestation","sign_citation":"https://pith.science/pith/TNXM66DT4Y7CU7ZIPYEMOPUCCM/action/citation_signature","submit_replication":"https://pith.science/pith/TNXM66DT4Y7CU7ZIPYEMOPUCCM/action/replication_record"}},"created_at":"2026-06-30T01:16:31.950230+00:00","updated_at":"2026-06-30T01:16:31.950230+00:00"}