{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:3BSYH5ML66C6PQIBFRZ4HXGBA3","short_pith_number":"pith:3BSYH5ML","schema_version":"1.0","canonical_sha256":"d86583f58bf785e7c1012c73c3dcc106d5dca1a9ee298848b42ad23b7e77393f","source":{"kind":"arxiv","id":"2605.15226","version":1},"attestation_state":"computed","paper":{"title":"Is Agentic AI Ready for Real-World Hardware Engineering? A Deep Dive with Phoenix-bench","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Software-tuned AI agents struggle with hardware engineering because bugs propagate through signal flows across instantiated modules rather than along call graphs.","cross_cats":["cs.AI","cs.SE"],"primary_cat":"cs.AR","authors_text":"Bingsheng He, Feng Yu, Hongshi Tan, Qingyun Zou, WengFai Wong","submitted_at":"2026-05-13T14:14:54Z","abstract_excerpt":"We ask whether agentic AI systems built for software engineering transfer to realistic hardware engineering. Existing hardware LLM benchmarks isolate sub-tasks but none jointly requires repository navigation, hierarchy-aware localization, Electronic Design Automation (EDA) executable verification, and maintenance-style patching. We introduce \\textbf{Phoenix-bench}, a synchronized corpus of 511 verified Verilator instances from 114 GitHub repositories, each shipped with the developer patch, design-flow labels, fail-to-pass and pass-to-pass testbenches, and a Docker-pinned EDA environment so res"},"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":true,"formal_links_present":true},"canonical_record":{"source":{"id":"2605.15226","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AR","submitted_at":"2026-05-13T14:14:54Z","cross_cats_sorted":["cs.AI","cs.SE"],"title_canon_sha256":"b260dbe540a2309f007647add5e11c061faa3163f40b17033c76fa95326e86d5","abstract_canon_sha256":"902eb9cd27c5ad8e290462ad99ed711299f34815d01cb4dc434a1f3b60727fdf"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:00:47.282913Z","signature_b64":"zDGosrOJOzRfQVuHIRga6JT3dp46eQnZV6B8k/EBfBqHaT9Y2Xuplp2X5DdC3Rlj5/botpWUsYjTSzpK/Jt8DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d86583f58bf785e7c1012c73c3dcc106d5dca1a9ee298848b42ad23b7e77393f","last_reissued_at":"2026-05-20T00:00:47.281949Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:00:47.281949Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Is Agentic AI Ready for Real-World Hardware Engineering? A Deep Dive with Phoenix-bench","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Software-tuned AI agents struggle with hardware engineering because bugs propagate through signal flows across instantiated modules rather than along call graphs.","cross_cats":["cs.AI","cs.SE"],"primary_cat":"cs.AR","authors_text":"Bingsheng He, Feng Yu, Hongshi Tan, Qingyun Zou, WengFai Wong","submitted_at":"2026-05-13T14:14:54Z","abstract_excerpt":"We ask whether agentic AI systems built for software engineering transfer to realistic hardware engineering. Existing hardware LLM benchmarks isolate sub-tasks but none jointly requires repository navigation, hierarchy-aware localization, Electronic Design Automation (EDA) executable verification, and maintenance-style patching. We introduce \\textbf{Phoenix-bench}, a synchronized corpus of 511 verified Verilator instances from 114 GitHub repositories, each shipped with the developer patch, design-flow labels, fail-to-pass and pass-to-pass testbenches, and a Docker-pinned EDA environment so res"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Software and hardware are fundamentally different engineering tasks: the same agent loses 37% to 58% from SWE-bench Verified to Phoenix-bench because hardware bugs propagate across parallel instantiated modules through signal flow rather than along a software-style call graph, and software-tuned agents stop at the symptom file instead of tracing back through the instantiation chain.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The 511 instances drawn from 114 GitHub repositories, together with their developer patches and testbenches, form a representative sample of real-world hardware engineering work that requires repository navigation, hierarchy-aware localization, EDA verification, and multi-file patching.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Phoenix-bench shows agentic AI systems lose 37-58% resolved rate when moving from SWE-bench Verified to hardware tasks because bugs spread across parallel modules via signal flow, with testbench feedback lifting performance by 42-45% while file-level oracles add only 1.4%.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Software-tuned AI agents struggle with hardware engineering because bugs propagate through signal flows across instantiated modules rather than along call graphs.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"d89d35a0c32a3b758f70b5086df730f87f6da4bc990be0dcfa7be1400611d24b"},"source":{"id":"2605.15226","kind":"arxiv","version":1},"verdict":{"id":"8abfb133-075e-407c-af20-a15097023b61","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T17:44:20.485864Z","strongest_claim":"Software and hardware are fundamentally different engineering tasks: the same agent loses 37% to 58% from SWE-bench Verified to Phoenix-bench because hardware bugs propagate across parallel instantiated modules through signal flow rather than along a software-style call graph, and software-tuned agents stop at the symptom file instead of tracing back through the instantiation chain.","one_line_summary":"Phoenix-bench shows agentic AI systems lose 37-58% resolved rate when moving from SWE-bench Verified to hardware tasks because bugs spread across parallel modules via signal flow, with testbench feedback lifting performance by 42-45% while file-level oracles add only 1.4%.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The 511 instances drawn from 114 GitHub repositories, together with their developer patches and testbenches, form a representative sample of real-world hardware engineering work that requires repository navigation, hierarchy-aware localization, EDA verification, and multi-file patching.","pith_extraction_headline":"Software-tuned AI agents struggle with hardware engineering because bugs propagate through signal flows across instantiated modules rather than along call graphs."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.15226/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"claim_evidence","ran_at":"2026-05-19T18:21:56.308771Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-19T18:01:18.618887Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T17:50:44.177290Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T13:33:22.832843Z","status":"skipped","version":"1.0.0","findings_count":0}],"snapshot_sha256":"8b140a2e460c256d1164c076defbc763e3d147bb08d1c900635a185cdf1a2a1b"},"references":{"count":50,"sample":[{"doi":"","year":2023,"title":"Benchmarking Large Language Models for Automated Verilog","work_id":"11975879-65f5-4e4b-bddb-516a96ee955c","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Thakur, Shailja and Ahmad, Baleegh and Pearce, Hammond and Tan, Benjamin and Dolan-Gavitt, Brendan and Karri, Ramesh and Garg, Siddharth , journal=. 2024 , publisher=","work_id":"ce7966ae-1d56-4f18-9bcb-082b11d1902a","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Liu, Mingjie and Pinckney, Nathaniel and Khailany, Brucek and Ren, Haoxing , booktitle=. 2023 , organization=","work_id":"308fcf02-ca47-4806-8082-858a6c895835","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Location is Key: Leveraging","work_id":"7f8b273a-72f1-4b32-9f69-4c67087a378b","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"arXiv preprint arXiv:2503.04057 , year=","work_id":"efc8abbe-e08b-457b-89b0-937b82ef3feb","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":50,"snapshot_sha256":"9e789ce38c49a54aa18d338a843644faa88af68ef992141d8924877ab05bb739","internal_anchors":7},"formal_canon":{"evidence_count":2,"snapshot_sha256":"a6aebab90429703aeca30a4c8cea08216e192d270908a40280888c22474556db"},"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":"2605.15226","created_at":"2026-05-20T00:00:47.282112+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.15226v1","created_at":"2026-05-20T00:00:47.282112+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.15226","created_at":"2026-05-20T00:00:47.282112+00:00"},{"alias_kind":"pith_short_12","alias_value":"3BSYH5ML66C6","created_at":"2026-05-20T00:00:47.282112+00:00"},{"alias_kind":"pith_short_16","alias_value":"3BSYH5ML66C6PQIB","created_at":"2026-05-20T00:00:47.282112+00:00"},{"alias_kind":"pith_short_8","alias_value":"3BSYH5ML","created_at":"2026-05-20T00:00:47.282112+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3BSYH5ML66C6PQIBFRZ4HXGBA3","json":"https://pith.science/pith/3BSYH5ML66C6PQIBFRZ4HXGBA3.json","graph_json":"https://pith.science/api/pith-number/3BSYH5ML66C6PQIBFRZ4HXGBA3/graph.json","events_json":"https://pith.science/api/pith-number/3BSYH5ML66C6PQIBFRZ4HXGBA3/events.json","paper":"https://pith.science/paper/3BSYH5ML"},"agent_actions":{"view_html":"https://pith.science/pith/3BSYH5ML66C6PQIBFRZ4HXGBA3","download_json":"https://pith.science/pith/3BSYH5ML66C6PQIBFRZ4HXGBA3.json","view_paper":"https://pith.science/paper/3BSYH5ML","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.15226&json=true","fetch_graph":"https://pith.science/api/pith-number/3BSYH5ML66C6PQIBFRZ4HXGBA3/graph.json","fetch_events":"https://pith.science/api/pith-number/3BSYH5ML66C6PQIBFRZ4HXGBA3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3BSYH5ML66C6PQIBFRZ4HXGBA3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3BSYH5ML66C6PQIBFRZ4HXGBA3/action/storage_attestation","attest_author":"https://pith.science/pith/3BSYH5ML66C6PQIBFRZ4HXGBA3/action/author_attestation","sign_citation":"https://pith.science/pith/3BSYH5ML66C6PQIBFRZ4HXGBA3/action/citation_signature","submit_replication":"https://pith.science/pith/3BSYH5ML66C6PQIBFRZ4HXGBA3/action/replication_record"}},"created_at":"2026-05-20T00:00:47.282112+00:00","updated_at":"2026-05-20T00:00:47.282112+00:00"}