{"record": {"aliases": ["Looks Right, Works Right: A Project-Level Benchmark for Multi-Screen Mobile App Generation"], "artifacts": [{"id": "arxiv:2607.28645", "kind": "paper", "url": "https://arxiv.org/abs/2607.28645"}, {"id": "gh:anoa12159-hue/mobileforge_eval", "kind": "repo", "url": "https://github.com/anoa12159-hue/mobileforge_eval"}], "authors": ["Fan Wu", "Cuiyun Gao", "Yiming Huang", "Yang Xiao", "Yujia Chen", "Qing Liao"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2607.28645", "source_url": "https://arxiv.org/abs/2607.28645"}, "categories": ["General AI", "Robotics & Embodied AI", "cs.HC"], "description": {"en": "A benchmark that tests whether multimodal LLMs can turn multi-screen mobile app designs into buildable projects with working navigation."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2607.28645", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2607.28645", "title": "MobileForge"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2607.28645", "modality": null, "name": "MobileForge", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"admission_policy_version": "2026-09-09.1", "confidence": "Medium", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-09-25T04:40:22Z", "data_status": "primary-source-indexed", "display_eligible": "true", "origin_record_id": "bm_looks-right-works-right-a-project-level-be_7041b757", "origin_source": "arxiv", "recognition_confidence": "0.7", "record_sha256": "560777e72d21d9209a720557f7c91b02ea5370e85fa6563ff7271a6114a7c1b0", "relation": "introduces", "review_model": "claude-code-default", "review_state": "ai-reviewed", "reviewed_at": "2026-09-24T16:50:36.102096Z", "source_url": "https://arxiv.org/abs/2607.28645"}, "publisher": null, "released": "2026-05-29", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2607.28645", "source_url": "https://arxiv.org/abs/2607.28645"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2607-28645", "source": "claire_radar", "source_benchmark_id": "2607.28645", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": ["Looks Right, Works Right: A Project-Level Benchmark for Multi-Screen Mobile App Generation"], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Robotics & Embodied AI", "attentionForecast": {"confidence": "Low", "horizon": "7d", "reason": "This is a newly released benchmark in the active design-to-code area. Its public repository could draw some early attention, but there are no adoption signals yet.", "score": 45}, "capabilities": [], "confidence": "Medium", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-24T16:50:36.102096Z", "inputHash": "e4fae54d6717d6da5415fb383f50018735bebae2e06f8e566cd7abf99b03905b", "model": "claude-code-default"}, "curation": {"admissionPolicyVersion": "2026-09-09.1", "canonicalNameEvidence": "We introduce MobileForge, the first benchmark for project-level multi-screen mobile app generation", "canonicalNameSource": "abstract", "decisionReason": "The abstract names the benchmark, defines a five-axis scoring protocol with navigation test specifications, and links a public evaluation repository.", "evaluationEvidence": {"reuse": "The benchmark and supporting materials are available at https://github.com/anoa12159-hue/mobileforge_eval", "scoring": "five-axis evaluation of build, navigation, visual fidelity, code maintainability, and efficiency", "task": "project-level multi-screen mobile app generation"}, "model": "claude-code-default", "reviewedAt": "2026-09-24T16:50:36.102096Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "A benchmark that tests whether multimodal LLMs can turn multi-screen mobile app designs into buildable projects with working navigation.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["benchmark term in abstract", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL"], "snippet": "We introduce MobileForge, the first benchmark for project-level multi-screen mobile app generation, comprising real mobile apps, human-reviewed screens, structured page-relationship annotations, and navigation test specifications."}, "familyId": "bmf_c3462c64c030", "firstSeenAt": "2026-08-19", "heat": null, "id": "bm_looks-right-works-right-a-project-level-be_7041b757", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/anoa12159-hue/mobileforge_eval", "data": null, "paper": "https://arxiv.org/abs/2607.28645", "pdf": "https://arxiv.org/pdf/2607.28645", "project": null, "report": "https://arxiv.org/abs/2607.28645"}, "metrics": [], "motivation": "Recent multimodal large language models can convert visual designs directly into executable code, but real mobile products require multiple screenshots to become a buildable codebase with shared components and working navigation.", "name": "MobileForge", "oneLine": "A benchmark that tests whether multimodal LLMs can turn multi-screen mobile app designs into buildable projects with working navigation.", "paperTitle": "Looks Right, Works Right: A Project-Level Benchmark for Multi-Screen Mobile App Generation", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2607.28645", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-08-19T11:05:13.395391Z"}, "publications": [], "publishers": [{"name": "anoa12159-hue", "organizationType": "community", "role": "benchmark-publisher", "sourceUrl": "https://github.com/anoa12159-hue/mobileforge_eval"}], "readiness": "Runnable", "recognitionConfidence": 0.7, "relation": "introduces", "releasedAt": "2026-05-29", "reviewLifecycle": {"state": "pending"}, "source": {"authors": ["Fan Wu", "Cuiyun Gao", "Yiming Huang", "Yang Xiao", "Yujia Chen", "Qing Liao"], "categories": ["cs.HC", "cs.AI"], "id": "2607.28645", "title": "Looks Right, Works Right: A Project-Level Benchmark for Multi-Screen Mobile App Generation", "type": "arxiv", "url": "https://arxiv.org/abs/2607.28645"}, "sourceUpdatedAt": "2026-05-29T05:48:19Z", "topics": ["cs.HC"], "venueAttempts": [], "whyItMatters": "It moves design-to-code evaluation from single pages to whole projects. It scores cross-page navigation and maintainability, which single-page benchmarks do not measure."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
