DRAFT-Bench 发布：评测前沿 Agent 在专业设计全流程中的可靠性
DRAFT-Bench: Evaluating Frontier Agents Across the Professional Design Lifecycle
一个新的设计类 Agent 基准，Astra 全跑完但违规多，Claude 少做几个却更稳，论文本身还很克制地说明评测局限。
论文提出 DRAFT-Bench（Design Reliability Across Feedback and Tasks），覆盖需求澄清、创意探索、约束编辑、迭代修改和生产交付五个设计环节。测试对比了 Codex、Astra 和 Claude Code，使用 35 份设计简报、43 个条件，另有 40 个场景的 RFQ 独立赛道。Astra 完成 43/43 个设计条件，Claude 完成 35/43；在 28 个共同评分的核心条件中，语义检查对 Astra 标记 11 项违规、Claude 5 项。论文指出这些是模型判断而非专家确认的失败率，并强调执行能力、需求合规与专业验收需要分开评估。
DRAFT-Bench: Evaluating Frontier Agents Across the Professional Design Lifecycle
Can frontier general-purpose agents understand client intent, explore creative directions, execute revisions, and prepare production deliverables? We introduce DRAFT-Bench (Design Reliability Across Feedback and Tasks), a lifecycle-oriented evaluation of requirement elicitation, creative exploration, constrained editing, iterative revision, and production handoff. Its Requirement-to-Artifact framework links evolving requirements, protected properties, artifact versions, and delivery obligations. We compare native Codex/Astra and Claude Code across 35 design briefs under 43 conditions, alongside a separate 40-scenario RFQ track. Astra completes 43/43 design conditions and Claude 35/43. All scored core runs receive full structural credit, yet on the 28 core conditions scored for both systems, semantic checks flag violations in 11 for Astra and 5 for Claude; these are model judgments, not expert-confirmed failure rates. RFQ coverage is high, but success ordering changes across cohorts. SKU expansion, revision dependencies, and White ink files expose ambiguity among unmet obligations, missing inputs, and evaluator interpretations. These findings separate execution, requirement compliance, and professional acceptance. Unpaired tracks and different tools preclude claims of RFQ-to-delivery success, factory acceptance, or a pure model ranking.