记录 SAGE、AI Agent 工程、后端系统与持续成长。https://blog.sagecompanion.top/https://blog.sagecompanion.top/posts/sage-from-question-to-growth/https://blog.sagecompanion.top/posts/sage-from-question-to-growth/SAGE 为什么不只是另一个聊天框，以及目标、知识、实践和证据如何形成可恢复的学习闭环。Fri, 24 Jul 2026 00:00:00 GMThttps://blog.sagecompanion.top/posts/chat-harness-2/https://blog.sagecompanion.top/posts/chat-harness-2/从事件流、上下文预算、持久化和工具审批出发，拆解一个可恢复 Agent Harness 的工程边界。Wed, 22 Jul 2026 00:00:00 GMThttps://blog.sagecompanion.top/posts/agent-memory-boundaries/https://blog.sagecompanion.top/posts/agent-memory-boundaries/Memory、Context 与 Knowledge 经常被混为一谈，工程上却需要不同的写入门禁和生命周期。Mon, 20 Jul 2026 00:00:00 GMThttps://blog.sagecompanion.top/posts/java-to-agent-engineering/https://blog.sagecompanion.top/posts/java-to-agent-engineering/Agent 工程不是抛弃后端基础，而是把协议、状态、权限和可观测性带进模型驱动系统。Sat, 18 Jul 2026 00:00:00 GMT