人工智能记忆管理技能——反思、碎片整理等
概览
基础记忆技能——教AI编码代理如何有效使用基础记忆的MCP工具。 技能是 AI 代理加载的 markdown 指令文件（SKILL.md），用于领域特定指导。每个技能包含关于何时以及如何使用特定工具的结构化知识，并附有示例和最佳实践。 Basic Memory 提供了 MCP 服务器——如 write_note、search_notes 和 build_context 等工具，用于管理本地优先的知识图谱。这些技能教会代理如何正确使用这些工具：何时创建任务或笔记，如何结构化观察以保证可搜索性，何时运行模式验证，等等。
概述（AI 翻译）
基础记忆技能——教AI编码代理如何有效使用基础记忆的MCP工具。
什么是技能？
技能是 Markdown 指令文件（[SKILL.md），AI 代理加载用于领域特定指导。每个技能包含关于何时以及如何使用特定工具的结构化知识，并附有示例和最佳实践。
Basic Memory 提供了 MCP 服务器——如 write_note、search_notes 和 build_context 等工具，用于管理本地优先的知识图谱。这些技能教会代理如何正确使用这些工具：何时创建任务与笔记，如何结构化可搜索的观测值，何时运行模式验证等。
技能
| 技能 | 描述 | 何时使用 |
|---|---|---|
| 内存任务 | 结构化任务跟踪，能够通过上下文压缩而存活。创建带有步骤、状态和上下文的类型Task笔记。 |
多步工作（3+步骤），任何可能超过上下文窗口或压缩后恢复的工作。 |
| 内存模式 | 模式生命周期管理——发现未模式化的笔记，推断模式，创建/编辑定义，验证并检测漂移。 |
本页由 AI 翻译，技术细节请以原文为准。
README
以下为原文 README 节选，完整内容请查看原文。
basic-memory-skills
Skills for Basic Memory — teach AI coding agents how to use Basic Memory’s MCP tools effectively.
What Are Skills?
Skills are markdown instruction files (SKILL.md) that AI agents load for domain-specific guidance. Each skill contains structured knowledge about when and how to use specific tools, with examples and best practices.
Basic Memory provides the MCP server — tools like write_note, search_notes, and build_context for managing a local-first knowledge graph. These skills teach agents how to use those tools well: when to create tasks vs. notes, how to structure observations for searchability, when to run schema validation, and more.
Skills
| Skill | Description | When to use |
|---|---|---|
| memory-tasks | Structured task tracking that survives context compaction. Creates typed Task notes with steps, status, and context. |
Multi-step work (3+ steps), anything that might outlast a context window, or after compaction to resume. |
| memory-schema | Schema lifecycle management — discover unschemaed notes, infer schemas, create/edit definitions, validate, and detect drift. | When structured note types emerge (Task, Person, Meeting, etc.) and you want consistency. |
| memory-reflect | Sleep-time memory reflection. Reviews recent conversations and daily notes, extracts insights, consolidates into long-term memory. Inspired by sleep-time compute. | Schedule via cron (1-2x daily), trigger from heartbeat, or run on demand. |
| memory-notes | How to write well-structured notes — frontmatter, observations with semantic categories, relations with wiki-links, and best practices. | When creating or improving notes, or when you need a reference for the note format. |
| memory-metadata-search | Structured metadata search — query notes by custom frontmatter fields using equality, range, array, and nested filters. | When finding notes by status, priority, confidence, or any custom YAML field. |
| memory-defrag | Memory defragmentation — split bloated files, merge duplicates, remove stale information, restructure the hierarchy. | Run weekly/biweekly via cron, or on demand when memory feels messy. |
| memory-lifecycle | Entity lifecycle management — status transitions through folder-based organization, archiving completed work. Core principle: archive, never delete. | When marking items complete, archiving old entities, or managing folder-based status workflows. |
| memory-ingest | Process unstructured external input into structured entities. Parses meeting transcripts, conversation logs, and pasted documents. | When pasting a transcript, conversation log, or external document that should become structured knowledge. |
推荐工具
换一个关键词，或者移除筛选条件。
安装
npx skillfish add basicmachines-co/basic-memory-skills