{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:G6UURBCV4AWTDN5LKMP4A4VPOJ","short_pith_number":"pith:G6UURBCV","schema_version":"1.0","canonical_sha256":"37a9488455e02d31b7ab531fc072af726924d4a8a5871f2784d1f450bf2f37df","source":{"kind":"arxiv","id":"2606.15673","version":2},"attestation_state":"computed","paper":{"title":"Where Did It Go Wrong? Process-Level Evaluation of Web Agents with Semantic State Tracking","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"JiHyuk Byun, Jiwan Chung, Seon Joo Kim, Vibhav Vineet","submitted_at":"2026-04-08T07:25:29Z","abstract_excerpt":"Web agents act through long interaction sequences, yet existing benchmarks evaluate only terminal success, discarding all process information and offering little guidance on improvement. In this work, we conduct a process-level analysis of web agents. We introduce WebStep, a benchmark of 1,800 task instances with controlled difficulty and automatic semantic state tracking. Each website exposes a deterministic semantic MDP alongside the GUI: the agent operates on the interface, while the environment records high-level states and transitions in the background, enabling fine-grained analysis with"},"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":"2606.15673","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-04-08T07:25:29Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"3bdeb27b4267d3ba3418b7c53dc610c8d4415f823d6a4d050674c81fffcaa74c","abstract_canon_sha256":"f944ad2f920baef0f3d55effd14eb56cca8b6370276a34394fc64866e5ed0af9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-05T00:42:06.700940Z","signature_b64":"lDcunNukUhhcrWo+ScPG8rFVGRKQO2NwUhxt8craQDsZu+lf7HVzJs+JHciKUQ24BGydh78E/cnaNVqwcJQ0Dw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"37a9488455e02d31b7ab531fc072af726924d4a8a5871f2784d1f450bf2f37df","last_reissued_at":"2026-08-05T00:42:06.698059Z","signature_status":"signed_v1","first_computed_at":"2026-08-05T00:42:06.698059Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Where Did It Go Wrong? Process-Level Evaluation of Web Agents with Semantic State Tracking","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"JiHyuk Byun, Jiwan Chung, Seon Joo Kim, Vibhav Vineet","submitted_at":"2026-04-08T07:25:29Z","abstract_excerpt":"Web agents act through long interaction sequences, yet existing benchmarks evaluate only terminal success, discarding all process information and offering little guidance on improvement. In this work, we conduct a process-level analysis of web agents. We introduce WebStep, a benchmark of 1,800 task instances with controlled difficulty and automatic semantic state tracking. Each website exposes a deterministic semantic MDP alongside the GUI: the agent operates on the interface, while the environment records high-level states and transitions in the background, enabling fine-grained analysis with"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.15673","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/2606.15673/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":"2606.15673","created_at":"2026-08-05T00:42:06.699130+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.15673v2","created_at":"2026-08-05T00:42:06.699130+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.15673","created_at":"2026-08-05T00:42:06.699130+00:00"},{"alias_kind":"pith_short_12","alias_value":"G6UURBCV4AWT","created_at":"2026-08-05T00:42:06.699130+00:00"},{"alias_kind":"pith_short_16","alias_value":"G6UURBCV4AWTDN5L","created_at":"2026-08-05T00:42:06.699130+00:00"},{"alias_kind":"pith_short_8","alias_value":"G6UURBCV","created_at":"2026-08-05T00:42:06.699130+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.26041","citing_title":"Desktop-Delta Bench: Do Computer-Use Models Understand Desktop GUI Transitions?","ref_index":23,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/G6UURBCV4AWTDN5LKMP4A4VPOJ","json":"https://pith.science/pith/G6UURBCV4AWTDN5LKMP4A4VPOJ.json","graph_json":"https://pith.science/api/pith-number/G6UURBCV4AWTDN5LKMP4A4VPOJ/graph.json","events_json":"https://pith.science/api/pith-number/G6UURBCV4AWTDN5LKMP4A4VPOJ/events.json","paper":"https://pith.science/paper/G6UURBCV"},"agent_actions":{"view_html":"https://pith.science/pith/G6UURBCV4AWTDN5LKMP4A4VPOJ","download_json":"https://pith.science/pith/G6UURBCV4AWTDN5LKMP4A4VPOJ.json","view_paper":"https://pith.science/paper/G6UURBCV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.15673&json=true","fetch_graph":"https://pith.science/api/pith-number/G6UURBCV4AWTDN5LKMP4A4VPOJ/graph.json","fetch_events":"https://pith.science/api/pith-number/G6UURBCV4AWTDN5LKMP4A4VPOJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/G6UURBCV4AWTDN5LKMP4A4VPOJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/G6UURBCV4AWTDN5LKMP4A4VPOJ/action/storage_attestation","attest_author":"https://pith.science/pith/G6UURBCV4AWTDN5LKMP4A4VPOJ/action/author_attestation","sign_citation":"https://pith.science/pith/G6UURBCV4AWTDN5LKMP4A4VPOJ/action/citation_signature","submit_replication":"https://pith.science/pith/G6UURBCV4AWTDN5LKMP4A4VPOJ/action/replication_record"}},"created_at":"2026-08-05T00:42:06.699130+00:00","updated_at":"2026-08-05T00:42:06.699130+00:00"}