{"paper":{"title":"LEGO: An LLM Skill-Based Front-End Design Generation Platform","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"Decomposing front-end design into six steps and 42 reusable circuit skills lets LLMs reach 80.5 percent success on hard RTL tasks that direct prompting cannot solve.","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Jiecheng Ma, Jincheng Lou, Runzhe Tao, Ruohan Xu, Xinyu Qu, Yibo Lin","submitted_at":"2026-04-25T15:44:48Z","abstract_excerpt":"Existing LLM-based EDA agents are often isolated task-specific systems. This leads to repeated engineering effort and limited reuse of successful design and debugging strategies. We present LEGO, a unified skill-based platform for front-end design generation. It decomposes the digital front-end flow into six independent steps and represents every agent capability as a standardized composable circuit skill within a plug-and-play architecture. To build this skill library, we survey more than 100 papers, select 11 representative open-source projects, and extract 42 executable circuit skills withi"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"individual circuit skills constructed within LEGO raise Pass@1 from 0.000 to 0.805. This is an 80.5% gain over the baseline. Cross-project skill compositions also reach 0.805 Pass@1. They outperform hierarchy-verilog by 14.6% and VerilogCoder by 2.5%.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the digital front-end flow decomposes cleanly into six independent steps and that the 42 skills extracted from 11 surveyed projects are representative, generalizable, and composable without significant performance loss on new problems.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"LEGO extracts 42 circuit skills from open-source projects to enable composable LLM-based front-end design, raising Pass@1 to 0.805 on challenging Verilog problems.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Decomposing front-end design into six steps and 42 reusable circuit skills lets LLMs reach 80.5 percent success on hard RTL tasks that direct prompting cannot solve.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"ebd3653e5c91c1165dff4bfb8d71d0b906b86830a426ed9a384755a8b78af88f"},"source":{"id":"2604.23355","kind":"arxiv","version":2},"verdict":{"id":"e75a2235-643e-44bb-bfa7-c990629424d9","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-08T08:17:29.646005Z","strongest_claim":"individual circuit skills constructed within LEGO raise Pass@1 from 0.000 to 0.805. This is an 80.5% gain over the baseline. Cross-project skill compositions also reach 0.805 Pass@1. They outperform hierarchy-verilog by 14.6% and VerilogCoder by 2.5%.","one_line_summary":"LEGO extracts 42 circuit skills from open-source projects to enable composable LLM-based front-end design, raising Pass@1 to 0.805 on challenging Verilog problems.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the digital front-end flow decomposes cleanly into six independent steps and that the 42 skills extracted from 11 surveyed projects are representative, generalizable, and composable without significant performance loss on new problems.","pith_extraction_headline":"Decomposing front-end design into six steps and 42 reusable circuit skills lets LLMs reach 80.5 percent success on hard RTL tasks that direct prompting cannot solve."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.23355/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_compliance","ran_at":"2026-05-19T23:14:59.539411Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"407b05c3085bd3d603f5ce4f643f32199d086df5aafa5f221416a38fdbcd3d8f"},"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"}