← Back to Skills Marketplace
renguanjie

elite memory skill

by renguanjie · GitHub ↗ · v1.0.1 · MIT-0
cross-platform ⚠ suspicious
301
Downloads
0
Stars
0
Active Installs
2
Versions
Install in OpenClaw
/install elite-memory-skill
Description
终极 AI 记忆系统 - WAL 协议 + 每日记忆 + 长期记忆 + GitHub 自动同步 + 飞书通知
README (SKILL.md)

Elite Memory - 终极 AI 记忆系统

为 AI Agent 打造的完整记忆管理系统,支持双层记忆架构、GitHub 自动同步、飞书通知。

核心功能

双层记忆架构

当日对话 → memory/YYYY-MM-DD-temp.md (临时记忆,实时写入)
              ↓
         次日 08:00
              ↓
    分析临时记忆 → 提取关键点
              ↓
    ┌───────────┴───────────┐
    ↓                       ↓
memory/YYYY-MM-DD.md   MEMORY.md
(正式日记)            (长期记忆提炼)

自动化任务

时间 任务 脚本
每日 23:55 Git 同步到 GitHub sync-memory-to-github.sh
每日 08:00 读取昨日记忆 + 创建今日记忆 morning-memory-read.sh
每次对话后 实时写入临时记忆 手动调用 analyze-memory.sh

使用方法

初始化记忆系统

# 1. 创建记忆目录
node scripts/init.mjs

# 2. 配置 GitHub remote
gh repo create ai-memory --private
git remote add memory [email protected]:username/ai-memory.git

# 3. 配置飞书通知 (可选)
export FEISHU_USER_ID="ou_xxxxx"

记忆操作

# 分析对话并写入临时记忆
node scripts/analyze.mjs "对话内容" --topic "主题"

# 手动同步到 GitHub
node scripts/sync.mjs

# 查看记忆状态
node scripts/status.mjs

定时任务配置

# 添加到 crontab
crontab -e

# 每日 23:55 同步
55 23 * * * ~/.openclaw/workspace/scripts/sync-memory-to-github.sh

# 每日 08:00 读取
0 8 * * * ~/.openclaw/workspace/scripts/morning-memory-read.sh

记忆文件结构

workspace/
├── MEMORY.md                      # 长期记忆 (curated)
├── SESSION-STATE.md               # 会话状态
└── memory/
    ├── YYYY-MM-DD-temp.md         # 临时记忆 (当日)
    ├── YYYY-MM-DD.md              # 正式记忆 (归档后)
    ├── projects/
    │   └── [project-name].md      # 项目记忆
    ├── skills/
    │   └── [skill-name].md        # 技能学习记忆
    └── people/
        └── [person-name].md       # 人员记忆

飞书通知

配置 FEISHU_USER_ID 环境变量后,同步完成会自动发送通知:

  • ✅ 同步成功:包含变更详情
  • ❌ 同步失败:包含错误信息

WAL 协议 (Write-Ahead Logging)

记忆写入规则:

  1. 用户表达偏好 → 立即写入 MEMORY.md → 再回应
  2. 用户做出决策 → 立即写入当日记忆 + MEMORY.md → 再回应
  3. 发现错误/教训 → 立即写入当日记忆 → 再回应
  4. 项目相关上下文 → 写入 memory/projects/[project].md

故障排查

日志位置

# 同步日志
tail -f ~/.openclaw/workspace/logs/memory-sync.log

# 晨间读取日志
tail -f ~/.openclaw/workspace/logs/memory-morning.log

常见问题

Q: Git 推送失败 A: 检查网络连接,运行 git pull --rebase memory main 后再试

Q: 飞书通知不发送 A: 检查 FEISHU_USER_ID 是否配置正确

Q: 记忆文件丢失 A: 从 GitHub 恢复:git pull memory main

相关资源

  • GitHub 仓库:https://github.com/renguanjie/ai-memory
  • 文档:https://docs.openclaw.ai/memory
Usage Guidance
This skill appears to be a local memory manager that writes daily memory files and can push them to a GitHub repo and notify a Feishu user. Before installing: 1) Inspect the package contents — SKILL.md references Node scripts (init.mjs, analyze.mjs, sync.mjs, status.mjs) that are not included; ask the publisher why they are missing or obtain the complete source. 2) Confirm you actually need 'node' on PATH — if the implementation is purely shell-based you can remove the node requirement. 3) Understand and control Git behavior: the scripts will git add/commit/push to whatever 'memory' remote you configure — do not configure a remote pointing to a public or sensitive repo until you’ve reviewed which files will be committed. 4) FEISHU_USER_ID is declared required in metadata but described as optional in docs — decide whether you want Feishu notifications and verify how the 'openclaw message' command will behave on your system. 5) If you install, review and run install.sh in a safe environment, and do not add the suggested crontab entries until you’ve tested the scripts manually. If the author can provide the missing Node scripts or a clear explanation for their absence, and clarify the FEISHU_USER_ID requirement, that would reduce concerns.
Capability Analysis
Type: OpenClaw Skill Name: elite-memory-skill Version: 1.0.1 The elite-memory-skill is a structured memory management system that archives AI agent interactions into daily logs and long-term memory files. It includes scripts for automated synchronization with a user-configured private GitHub repository and optional Feishu notifications. The code logic in scripts like `sync-memory-to-github.sh` and `analyze-memory.sh` is consistent with the stated purpose of maintaining a persistent, version-controlled history of agent state and user preferences without any evidence of malicious intent or unauthorized data exfiltration.
Capability Assessment
Purpose & Capability
The skill claims a Node-based workflow (SKILL.md shows node scripts: init.mjs, analyze.mjs, sync.mjs, status.mjs) and declares 'node' as a required binary, but the package only includes Bash scripts (analyze-memory.sh, sync-memory-to-github.sh, morning-memory-read.sh). Requiring 'node' and advertising Node scripts without those files is inconsistent and could indicate missing/altered files or sloppy packaging. FEISHU_USER_ID is declared required in metadata but SKILL.md describes it as optional — another mismatch.
Instruction Scope
Runtime instructions and the included shell scripts read and write files under ~/.openclaw/workspace (memory files, SESSION-STATE.md, MEMORY.md), run git operations (add/commit/push/pull) and attempt to send notifications via an 'openclaw' command to Feishu using FEISHU_USER_ID. Those actions are coherent with a memory-sync skill, but the SKILL.md instructs users to run Node scripts that are not present. The scripts will push local data to a Git remote you must configure — that can leak personal data to GitHub if misconfigured, so the push behavior should be explicitly reviewed before use.
Install Mechanism
No remote downloads or package installs. install.sh simply copies local files into the target workspace and makes script files executable. This is low-risk compared to installs that fetch arbitrary archives from the network.
Credentials
The only declared environment requirement is FEISHU_USER_ID (primaryEnv), which aligns with Feishu notifications, but SKILL.md treats Feishu config as optional. 'git' is required and used; 'node' is declared required but not used by the provided scripts — disproportionate. There's no request for unrelated credentials, but you should verify whether FEISHU_USER_ID must be set for operation and why node is listed.
Persistence & Privilege
The skill is not set to always:true and does not request persistent platform-wide privileges. Installation copies files into the user's workspace and suggests adding cron entries; it does not modify other skills or system-wide agent settings.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install elite-memory-skill
  3. After installation, invoke the skill by name or use /elite-memory-skill
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.1
- No user-visible changes in this release. - Version bump only; documentation, features, and functionality remain unchanged.
v1.0.0
Elite Memory Skill 1.0.0 – Initial Release - Implements a two-tier memory architecture: daily (temporary and formal) and long-term storage. - Enables automated GitHub sync and Feishu notifications for memory events. - Provides scripts for memory analysis, synchronization, status checks, and initialization. - Integrates scheduled tasks for daily sync and memory review. - Documents memory structure, operations, WAL write protocol, troubleshooting, and setup instructions.
Metadata
Slug elite-memory-skill
Version 1.0.1
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 2
Frequently Asked Questions

What is elite memory skill?

终极 AI 记忆系统 - WAL 协议 + 每日记忆 + 长期记忆 + GitHub 自动同步 + 飞书通知. It is an AI Agent Skill for Claude Code / OpenClaw, with 301 downloads so far.

How do I install elite memory skill?

Run "/install elite-memory-skill" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is elite memory skill free?

Yes, elite memory skill is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does elite memory skill support?

elite memory skill is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created elite memory skill?

It is built and maintained by renguanjie (@renguanjie); the current version is v1.0.1.

💬 Comments