c-memory
关于
c-memory enables persistent memory across Claude Code sessions by storing facts, preferences, and context in local markdown files. It automatically loads key information from `~/.claude/memory/MEMORY.md` at session start and allows saving important learnings during interactions. Developers should use this skill to maintain continuity and personalization across multiple coding sessions.
快速安装
Claude Code
推荐npx skills add daxaur/openpaw -a claude-code/plugin add https://github.com/daxaur/openpawgit clone https://github.com/daxaur/openpaw.git ~/.claude/skills/c-memory在 Claude Code 中复制并粘贴此命令以安装该技能
技能文档
Memory — Persistent Context
Store and recall facts, preferences, and context across Claude Code sessions.
How It Works
Memory lives in ~/.claude/memory/ as plain markdown files. Read them at session start, write to them when you learn something important.
Directory Structure
~/.claude/memory/
├── MEMORY.md # Key facts — always read this first
├── people.md # People the user mentions (names, roles, preferences)
├── preferences.md # User preferences (tools, workflows, habits)
├── projects.md # Active projects and their context
└── journal.md # Session log — append-only, dated entries
Session Start
At the beginning of every session, read ~/.claude/memory/MEMORY.md to load context. This file should be concise — under 100 lines.
When to Save
Save to memory when the user:
- Shares their name, role, or preferences
- Mentions people they work with
- Describes a project or recurring task
- Says "remember this" or "don't forget"
- Corrects you about something
Writing Rules
- MEMORY.md: Key facts only. Keep under 100 lines. Update, don't append.
- people.md: Name, role, relationship, preferences. One section per person.
- preferences.md: Tools, workflows, communication preferences.
- projects.md: Project name, status, key files, context.
- journal.md: Append a dated entry at the end of each session summarizing what was done.
Commands
# Read memory
cat ~/.claude/memory/MEMORY.md
# List memory files
ls ~/.claude/memory/
# Append to journal
echo "## $(date +%Y-%m-%d)" >> ~/.claude/memory/journal.md
Example MEMORY.md
# Memory
## User
- Name: Alex
- Role: Frontend developer
- Prefers TypeScript, uses Neovim
- Timezone: PST
## Active Projects
- ctrl.build — DeFi workflow automation (Next.js + Solidity)
- openpaw — CLI tool for Claude Code setup
## Preferences
- Concise responses, no emoji unless asked
- Always use bun instead of npm
- Dark mode everything
Guidelines
- Never store passwords, API keys, or secrets in memory files
- Always ask before overwriting existing memory entries
- If memory files don't exist, create them on first write
- Keep MEMORY.md focused — move details to topic-specific files
Obsidian Sync (Auto)
If c-obsidian is also installed (check ls ~/.claude/skills/c-obsidian/), automatically sync memory to Obsidian:
On Memory Write
When saving to any memory file, also mirror to Obsidian in an AI/ folder:
obsidian-cli create "AI/Memory" --content "$(cat ~/.claude/memory/MEMORY.md)"
obsidian-cli create "AI/People" --content "$(cat ~/.claude/memory/people.md)"
obsidian-cli create "AI/Preferences" --content "$(cat ~/.claude/memory/preferences.md)"
obsidian-cli create "AI/Projects" --content "$(cat ~/.claude/memory/projects.md)"
Journal → Daily Note
Append session logs to both journal.md and Obsidian's daily note:
obsidian-cli append "$(date +%Y-%m-%d)" "### Claude Session $(date +%H:%M)\n- [summary]"
echo "## $(date +%Y-%m-%d %H:%M)\n- [summary]" >> ~/.claude/memory/journal.md
On Session Start
After reading ~/.claude/memory/MEMORY.md, check Obsidian for newer content:
obsidian-cli search --folder "AI" "Memory"
If the Obsidian version has additional facts, merge them into ~/.claude/memory/MEMORY.md.
Conflict Resolution
~/.claude/memory/is the authoritative quick-access cache- Obsidian is the long-term archive and rich knowledge base
- When in doubt, prefer the most recently modified version
GitHub 仓库
相关推荐技能
qmd
开发这是一个本地搜索和索引的CLI工具,支持BM25、向量搜索和重排序功能。开发者可以用它快速索引本地文件(如Markdown文档)并进行混合搜索,特别适合代码库或文档的本地检索。它还提供MCP模式,能轻松集成到Claude开发环境中使用。
subagent-driven-development
开发该Skill用于在当前会话中执行包含独立任务的实施计划,它会为每个任务分派一个全新的子代理并在任务间进行代码审查。这种"全新子代理+任务间审查"的模式既能保障代码质量,又能实现快速迭代。适合需要在当前会话中连续执行独立任务,并希望在每个任务后都有质量把关的开发场景。
mcporter
开发mcporter Skill 让开发者能在Claude中直接管理和调用MCP服务器。它支持列出可用服务器、调用工具、处理OAuth认证以及管理服务器守护进程。开发者可以通过命令行式交互快速执行`mcporter list`查看服务器,或使用`mcporter call`直接调用工具,简化了MCP工作流程。
adk-deployment-specialist
开发这是一个用于部署和编排Google Vertex AI ADK智能体的Claude Skill,专为构建生产级多智能体系统而设计。它支持通过A2A协议进行智能体通信,提供代码执行沙箱和记忆库功能,并能处理智能体发现与任务提交。当开发者需要部署ADK智能体或编排多智能体协作时,可使用此Skill来简化Vertex AI Agent Engine的部署流程。
