Pai 的 AI 雷达
今天：Usage metrics API，AgentValley（Agent 原型）。并行还有 Bloblings、Fancy harnesses make、Anthropic’s new guide。
The Copilot usage metrics API now reports model usage from more of the tools your developers use with Copilot, including tools like OpenCode that weren’t previously captured in your reports. A new totals_by_client array is available in the enterprise, organization, enterprise-use
Every Claude Code agent, at a glance, in your Mac's notch Discussion | Link
A pet for every Claude Code session Discussion | Link
A controlled study shows that switching between coding-agent harnesses like Claude Code, OpenCode, or mini-SWE-agent barely moves pass rates more than rerunning the same setup. Cost per task, however, can triple, driven mainly by the system prompt that gets resent with every step
Anthropic published an 18-minute guide on Oct 22 called "When to fold subagents into your main agent loop." It is aimed at enterprise agent builders, but the lessons apply to anyone who set up a pile of subagents in Claude Code. Main points: - Default to one model, one context wi