Agent-Callable Feature Coverage：衡量软件对 AI 智能体的就绪度
原标题：Agent-Callable Feature Coverage: Measuring Software Readiness for AI Agents
AI 摘要
该论文提出「GUI–API 对齐原则」，即人类通过图形界面可用的能力也应通过结构化可调用接口向 AI 智能体开放。作者提出 Agent-Callable Feature Coverage（ACFC）产品级就绪度指标，以及基于 Accessibility、Discoverability、Controllability 三维度的 Agent Readiness Conformance（ARC）评估框架。对 30 个软件系统的实证研究发现，多数系统 API 覆盖中等，但在面向智能体的文档与安全治理上明显滞后；在 10 个系统的本地测试中，API-only 智能体完成 58 个任务中的 56 个，文档消融后降至 50 个，说明可发现性影响智能体执行。
正文节选
Agent-Callable Feature Coverage: Measuring Software Readiness for AI Agents Abstract AI agents already interact with graphical software through screenshot-based computer use; no GUI is truly off-limits anymore. The urgent question is therefore not whether agents can operate software, but how well software supports them through structured, controllable channels. We formalize this need as the GUI–API parity principle: every capability accessible to human users through a graphical interface should