{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:DCTL5LD4QHCD76QT6YQUKCKYPB","short_pith_number":"pith:DCTL5LD4","schema_version":"1.0","canonical_sha256":"18a6beac7c81c43ffa13f6214509587852924f54fb17618048cd69c53de5a11a","source":{"kind":"arxiv","id":"2605.26321","version":1},"attestation_state":"computed","paper":{"title":"Anchor: Mitigating Artifact Drift in Agent Benchmark Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Abhijay Rana, Maksim Ivanov","submitted_at":"2026-05-25T20:44:17Z","abstract_excerpt":"AI agents are beginning to complete valuable, long-horizon business operations tasks, but training and evaluation environments for enterprise work still struggle to balance realism, verifiability, and scale. Environment and task creation frequently suffers from a failure mode we call artifact drift: when instructions, environments, oracles, and verifiers are created by loosely coupled processes, they frequently disagree on what a task requires, producing environments that are unsolvable, reward-hackable, or inconsistent. We introduce Anchor, a task-generation pipeline that formalizes domain ex"},"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":"2605.26321","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-25T20:44:17Z","cross_cats_sorted":[],"title_canon_sha256":"25168af958d1ed1a39a597982626680bdd9341a22255c37c876a3d5f995fd53e","abstract_canon_sha256":"b56afa27aa3fddcee1a6e294ecf9bb73ff5f68c9f92157859d97e8e7b9066ff7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-27T01:05:11.925486Z","signature_b64":"pZ1YfkEFAZF6qmhymQIzOHn0nZcDLGB+kDuCZ0IVL2pNb/DGn/vH5DvLIBHNHt1Dmsgm310OlMXRYXVDJuXGBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"18a6beac7c81c43ffa13f6214509587852924f54fb17618048cd69c53de5a11a","last_reissued_at":"2026-05-27T01:05:11.924821Z","signature_status":"signed_v1","first_computed_at":"2026-05-27T01:05:11.924821Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Anchor: Mitigating Artifact Drift in Agent Benchmark Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Abhijay Rana, Maksim Ivanov","submitted_at":"2026-05-25T20:44:17Z","abstract_excerpt":"AI agents are beginning to complete valuable, long-horizon business operations tasks, but training and evaluation environments for enterprise work still struggle to balance realism, verifiability, and scale. Environment and task creation frequently suffers from a failure mode we call artifact drift: when instructions, environments, oracles, and verifiers are created by loosely coupled processes, they frequently disagree on what a task requires, producing environments that are unsolvable, reward-hackable, or inconsistent. We introduce Anchor, a task-generation pipeline that formalizes domain ex"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.26321","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/2605.26321/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":"2605.26321","created_at":"2026-05-27T01:05:11.924943+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.26321v1","created_at":"2026-05-27T01:05:11.924943+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.26321","created_at":"2026-05-27T01:05:11.924943+00:00"},{"alias_kind":"pith_short_12","alias_value":"DCTL5LD4QHCD","created_at":"2026-05-27T01:05:11.924943+00:00"},{"alias_kind":"pith_short_16","alias_value":"DCTL5LD4QHCD76QT","created_at":"2026-05-27T01:05:11.924943+00:00"},{"alias_kind":"pith_short_8","alias_value":"DCTL5LD4","created_at":"2026-05-27T01:05:11.924943+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/DCTL5LD4QHCD76QT6YQUKCKYPB","json":"https://pith.science/pith/DCTL5LD4QHCD76QT6YQUKCKYPB.json","graph_json":"https://pith.science/api/pith-number/DCTL5LD4QHCD76QT6YQUKCKYPB/graph.json","events_json":"https://pith.science/api/pith-number/DCTL5LD4QHCD76QT6YQUKCKYPB/events.json","paper":"https://pith.science/paper/DCTL5LD4"},"agent_actions":{"view_html":"https://pith.science/pith/DCTL5LD4QHCD76QT6YQUKCKYPB","download_json":"https://pith.science/pith/DCTL5LD4QHCD76QT6YQUKCKYPB.json","view_paper":"https://pith.science/paper/DCTL5LD4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.26321&json=true","fetch_graph":"https://pith.science/api/pith-number/DCTL5LD4QHCD76QT6YQUKCKYPB/graph.json","fetch_events":"https://pith.science/api/pith-number/DCTL5LD4QHCD76QT6YQUKCKYPB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DCTL5LD4QHCD76QT6YQUKCKYPB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DCTL5LD4QHCD76QT6YQUKCKYPB/action/storage_attestation","attest_author":"https://pith.science/pith/DCTL5LD4QHCD76QT6YQUKCKYPB/action/author_attestation","sign_citation":"https://pith.science/pith/DCTL5LD4QHCD76QT6YQUKCKYPB/action/citation_signature","submit_replication":"https://pith.science/pith/DCTL5LD4QHCD76QT6YQUKCKYPB/action/replication_record"}},"created_at":"2026-05-27T01:05:11.924943+00:00","updated_at":"2026-05-27T01:05:11.924943+00:00"}