{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:LQN7OZYFQ3J6EEXQVKMBKWA3UB","short_pith_number":"pith:LQN7OZYF","schema_version":"1.0","canonical_sha256":"5c1bf7670586d3e212f0aa9815581ba04d87d3895b9b467ac07320de2f29e80e","source":{"kind":"arxiv","id":"2608.04719","version":1},"attestation_state":"computed","paper":{"title":"Diagnosing Tool-Selection Reasoning in LLM Agents with Canary Tools","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Atul Anand, Sourav Chattaraj","submitted_at":"2026-08-05T11:38:33Z","abstract_excerpt":"Agent evaluations tell us that a model picked the wrong tool, but rarely why. We introduce canary tools: diagnostic probe tools planted in an agent's Model Context Protocol (MCP) tool set, each engineered to probe one specific tool-selection weakness. A six-type taxonomy (semantic decoys, parameter traps, capability mirages, prerequisite blindness, temporal decoys, and granularity traps) turns a single \"wrong tool\" outcome into a multi-dimensional profile of how a model reasons about tools. We evaluate eight models -- six hosted and two 8B open-weight -- spanning three capability tiers, on 120"},"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":"2608.04719","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-08-05T11:38:33Z","cross_cats_sorted":[],"title_canon_sha256":"30856efd3df7f0a82977db49a22c352202e8bb1d6d26fbee25cfc3ef7a531b67","abstract_canon_sha256":"9947783645084c657fd23bf53f393c287915e656c5361aeaad796672d76a10cd"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-06T01:46:35.964752Z","signature_b64":"bQbqdtk245lEvtwtkQNebAV+Qj5xBCmJFARNA6Jqe2xTiZfwJy5xSE6xY5hXAoUvPtXa7VncN6L5ifk9K+DZBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5c1bf7670586d3e212f0aa9815581ba04d87d3895b9b467ac07320de2f29e80e","last_reissued_at":"2026-08-06T01:46:35.963079Z","signature_status":"signed_v1","first_computed_at":"2026-08-06T01:46:35.963079Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Diagnosing Tool-Selection Reasoning in LLM Agents with Canary Tools","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Atul Anand, Sourav Chattaraj","submitted_at":"2026-08-05T11:38:33Z","abstract_excerpt":"Agent evaluations tell us that a model picked the wrong tool, but rarely why. We introduce canary tools: diagnostic probe tools planted in an agent's Model Context Protocol (MCP) tool set, each engineered to probe one specific tool-selection weakness. A six-type taxonomy (semantic decoys, parameter traps, capability mirages, prerequisite blindness, temporal decoys, and granularity traps) turns a single \"wrong tool\" outcome into a multi-dimensional profile of how a model reasons about tools. We evaluate eight models -- six hosted and two 8B open-weight -- spanning three capability tiers, on 120"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.04719","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/2608.04719/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":"2608.04719","created_at":"2026-08-06T01:46:35.965045+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.04719v1","created_at":"2026-08-06T01:46:35.965045+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.04719","created_at":"2026-08-06T01:46:35.965045+00:00"},{"alias_kind":"pith_short_12","alias_value":"LQN7OZYFQ3J6","created_at":"2026-08-06T01:46:35.965045+00:00"},{"alias_kind":"pith_short_16","alias_value":"LQN7OZYFQ3J6EEXQ","created_at":"2026-08-06T01:46:35.965045+00:00"},{"alias_kind":"pith_short_8","alias_value":"LQN7OZYF","created_at":"2026-08-06T01:46:35.965045+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/LQN7OZYFQ3J6EEXQVKMBKWA3UB","json":"https://pith.science/pith/LQN7OZYFQ3J6EEXQVKMBKWA3UB.json","graph_json":"https://pith.science/api/pith-number/LQN7OZYFQ3J6EEXQVKMBKWA3UB/graph.json","events_json":"https://pith.science/api/pith-number/LQN7OZYFQ3J6EEXQVKMBKWA3UB/events.json","paper":"https://pith.science/paper/LQN7OZYF"},"agent_actions":{"view_html":"https://pith.science/pith/LQN7OZYFQ3J6EEXQVKMBKWA3UB","download_json":"https://pith.science/pith/LQN7OZYFQ3J6EEXQVKMBKWA3UB.json","view_paper":"https://pith.science/paper/LQN7OZYF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.04719&json=true","fetch_graph":"https://pith.science/api/pith-number/LQN7OZYFQ3J6EEXQVKMBKWA3UB/graph.json","fetch_events":"https://pith.science/api/pith-number/LQN7OZYFQ3J6EEXQVKMBKWA3UB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LQN7OZYFQ3J6EEXQVKMBKWA3UB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LQN7OZYFQ3J6EEXQVKMBKWA3UB/action/storage_attestation","attest_author":"https://pith.science/pith/LQN7OZYFQ3J6EEXQVKMBKWA3UB/action/author_attestation","sign_citation":"https://pith.science/pith/LQN7OZYFQ3J6EEXQVKMBKWA3UB/action/citation_signature","submit_replication":"https://pith.science/pith/LQN7OZYFQ3J6EEXQVKMBKWA3UB/action/replication_record"}},"created_at":"2026-08-06T01:46:35.965045+00:00","updated_at":"2026-08-06T01:46:35.965045+00:00"}