OKF Agent Memory：为AI编码代理提供Git原生持久化内存系统

OKF Agent Memory是一套基于开放知识格式（OKF）v0.2标准的Git原生持久化项目内存系统，用于解决AI编码代理在上下文窗口关闭后的信息丢失问题1。该项目采用标准化、厂商中立的方案，将知识存储为Git仓库中的Markdown文件1。

系统采用多项优化技术以提升性能与可用性1。搜索性能可达300微秒以下，图验证耗时约4毫秒1。项目提供零依赖的Go工具链，命令行启动时间不超过5毫秒1。核心索引机制使用BM25本地词法索引，消除了向量嵌入API的成本1。此外，系统内置Model Context Protocol（MCP）服务器支持1，并通过Progressive Disclosure实现分层加载1。该项目采用MIT License开源许可证1。

OKF Agent Memory is a Git-native persistent memory system built on the Open Knowledge Format (OKF) v0.2 standard, designed to address the loss of information when AI coding agents' context windows close 1. The system adopts a standardized, vendor-neutral approach by storing knowledge as Markdown files within Git repositories 1.

The project delivers millisecond-level search performance, with searches completing in under 300 microseconds and graph verification taking approximately 4 milliseconds 1. It includes a zero-dependency Go toolchain with CLI startup times under 5 milliseconds and integrates a Model Context Protocol (MCP) server 1. Rather than relying on vector embedding APIs, the system employs BM25 local lexical indexing to eliminate associated costs 1. The solution also supports Progressive Disclosure for hierarchical data loading 1 and is distributed under the MIT License 1.