Overview
pmb saves project decisions, experience, and task facts to a local SQLite file so agents keep work context across sessions. It connects to many coding agents via MCP, auto-recalls relevant memory before model reasoning, and records new info after tasks end. Local-first architecture needs no cloud service or API key—fit for developers who care about data control and long-term project continuity. [Product features] Auto memory recall: match project experience, decisions, and file info before each chat. Cross-tool shared context: Claude Code, Cursor, Codex, Zed, and more share one memory store. Hybrid retrieval: keyword, vector, and entity relations rank memories. Memory quality management: from real agent execution, mark experience rules as valid, invalid, or to-verify. Visual memory map: relationship graphs and timelines for project facts, commits, and decision evolution. Local offline: data stays on the user's device—no telemetry, no cloud dependency.
Connected profiles
Sources · 0Sources & provenance
English text is maintained separately from the original record. Consult the original source before citing a fact.
No source links are available for this profile.
Original source text
Names may use phonetic renderings when no English name is supplied. The original wording is retained below.
pmb 将项目决策、经验、任务事实等信息保存至本地 SQLite 文件，让智能体跨会话保留工作上下文。它可通过 MCP 接入多种编程智能体，在模型推理前自动召回相关记忆，并在任务结束后记录新信息。产品采用本地优先架构，不依赖云端服务或 API 密钥，适合重视数据控制与长期项目连续性的开发者。【产品功能】自动记忆召回：在每次对话前匹配项目经验、决策与文件信息。跨工具共享上下文：支持 Claude Code、Cursor、Codex、Zed 等工具共用同一记忆库。混合检索机制：结合关键词、向量与实体关系进行记忆排序。记忆质量管理：根据智能体实际执行情况识别有效、无效或待验证的经验规则。可视化记忆地图：通过关系图与时间线查看项目事实、提交与决策演变。本地离线运行：数据保存在用户设备中，无遥测、无云端依赖。