Awublack Personal Memory System

一个智能化、高可靠性的个人知识操作系统。它不仅自动同步你的 MEMORY.md 到 SQLite 数据库并提供自然语言查询，更通过 Git 版本控制、守护进程和日志监控，确保你的记忆**永不丢失、可追溯、可恢复**。 **使用场景**： (1) 用户需要查询“我过去对 AI 模型的看法？” (2) 用户希望系统自...

Security Scan

The skill mostly does what it says (local Markdown→SQLite sync and local queries) but its instructions claim 'local-only' while also depending on a Git sync daemon and a remote GitHub repo — an incoherence that could lead to unintended data exfiltration or missing credential requirements.

Purpose & Capability

Name/description match the included code: auto_sync_memory.py converts MEMORY.md into memory.db and query_memory.py answers NL queries. However the SKILL.md hardcodes user-specific paths (/home/awu/...) and requires an Obsidian vault and a git-sync daemon; hardcoding a specific home path reduces portability and suggests this package was written for a single user environment rather than general use.

Instruction Scope

SKILL.md repeatedly states '完全本地化 / 不调用任何外部 API' yet also declares a remote Git repo (https://github.com/awublack/obsidian-vault) and requires a background git_sync_on_save.sh that '自动执行 git add/commit/push'. The instructions therefore direct periodic network pushes of MEMORY.md/obsidian-vault contents even though the doc claims no external calls. The skill also instructs running a background daemon with no guidance about what the daemon does or how credentials are provided — this is scope creep from 'local query' to automated network sync.

Install Mechanism

No install spec (instruction-only) and no downloads — the lowest install risk. The package includes two small Python scripts; nothing fetched from remote registries or arbitrary URLs during install.

Credentials

requires.env is empty, yet the documented workflow implicitly requires git credentials/SSH keys or HTTP credentials to push to the declared remote repo. The skill does not declare or request these credentials, which is an inconsistency: automatic Git pushes will either fail or proceed using the environment's existing Git credentials (silent exfiltration risk if credentials are present). Also requiring files under a specific /home/awu path is unusual for a general skill and may leak assumptions about where sensitive files live.

Persistence & Privilege

always:false (good). The skill recommends running a background daemon (git_sync_on_save.sh) that should be kept alive via nohup; running a persistent process has higher operational impact than a purely on-demand script. Autonomous invocation is allowed (platform default) — combine that with an automated push daemon and you have a continuous data flow to a remote if configured.

What to consider before installing

This package largely does what it says (sync MEMORY.md → memory.db and allow NL queries), but there are important inconsistencies you should verify before installing: - Inspect git_sync_on_save.sh before running it. The skill depends on that daemon to commit/push changes; if it's configured to push to a remote you must confirm the remote URL and that you intend to upload private notes. - SKILL.md claims 'local-only' but documents a remote Git repository. If you want purely local storage, disable or remove the git daemon, or change the git remote to a local-only location. - The skill uses hardcoded paths under /home/awu/.openclaw/workspace-work/. If your environment differs, update paths or run it in a sandbox. Hardcoded paths can cause the skill to miss data or operate on unexpected files. - The package does not declare any required credentials, yet git push requires credentials (SSH key or token). Determine which credentials Git will use in your environment and whether you want that behavior. - Review auto_sync_memory.py behavior: it deletes the memories table and repopulates it on sync (possible data-loss or overwrite of created_at). Make backups before first run. If you cannot review the git_sync_on_save.sh and the Git remote, treat this skill as potentially exfiltrating sensitive notes and test it in an isolated environment first.

Personal Memory System

概述

这是一个为 OpenClaw 设计的个人知识管理技能。它解决了“人类记忆不可靠”的核心问题，通过自动化流程，将你最重要的想法和决策从易遗忘的 Markdown 文本，转化为机器可查询的结构化数据库。

工作原理

1. 输入：你将重要的洞察、决策、经验写入 /home/awu/.openclaw/workspace-work/MEMORY.md 文件，使用 ### 标题分隔不同主题。
2. 同步：每当 OpenClaw 的心跳（heartbeat）被触发，auto_sync_memory.py 脚本会自动运行。
3. 处理：脚本读取 MEMORY.md，将每个 ### 标题及其下的内容解析为一条数据库记录。
4. 输出：这些记录被写入 /home/awu/.openclaw/workspace-work/memory.db SQLite 数据库。
5. 查询：当你提出自然语言问题（如“我过去对 AI 模型的看法？”），AI 助手会调用 query_memory.py 脚本，该脚本会将问题转换为 SQL 查询，从 memory.db 中精准检索出所有相关记忆，并为你生成一个自然语言的总结性回答。

系统组件

• MEMORY.md：你的个人长期记忆库，人类可读，可版本控制。
• memory.db：自动同步的 SQLite 数据库，机器可查询。
• auto_sync_memory.py：核心同步脚本，由 OpenClaw 心跳机制调用。
• query_memory.py：智能查询脚本，负责将自然语言问题转换为 SQL 查询并生成回答。
• obsidian-vault/：你的 Obsidian 知识库，所有笔记的原始来源。
• git_sync_on_save.sh：后台守护进程，每10秒检查 Obsidian 知识库变更，自动执行 git add/commit/push。
• git_sync.log：记录所有 Git 同步事件的日志文件，用于审计和故障排查。

安装与使用

1. 安装：将此技能包放入 OpenClaw 的 skills 目录，或通过 clawhub install awublack-personal-memory-system 安装。
2. 初始化：确保你的工作区根目录下存在以下文件和目录：
   • MEMORY.md
   • obsidian-vault/
   • git_sync_on_save.sh
   • git_sync.log
   • auto_sync_memory.py
   • query_memory.py
3. 运行：确保 git_sync_on_save.sh 已通过 nohup ./git_sync_on_save.sh > git_sync.log 2>&1 & 启动并常驻后台。
   • 检查：ps aux | grep git_sync_on_save.sh
   • 重启：pkill -f git_sync_on_save.sh && nohup ./git_sync_on_save.sh > git_sync.log 2>&1 &
4. 查询：直接向 AI 助手提问，例如：“我过去对 Obsidian 的看法是什么？” 系统会自动调用 query_memory.py 脚本检索并回答。

安全与隐私

• 完全本地化：所有数据（MEMORY.md, memory.db, query_memory.py, obsidian-vault/, git_sync_on_save.sh, git_sync.log）均存储在你的本地工作区，不上传、不共享。
• 无外部依赖：不调用任何网络 API，不访问云服务。
• 可审计：所有操作均在你的控制之下，你可以随时查看所有相关文件的内容。
• 三重冗余：你的记忆存在于三个独立位置：
  1. 本地工作区：MEMORY.md, memory.db, git_sync_on_save.sh
  2. Obsidian 知识库：/home/awu/.openclaw/workspace-work/obsidian-vault/
  3. 远程 Git 仓库：https://github.com/awublack/obsidian-vault 任何一处损坏，其他两处均可恢复，记忆永不丢失。

未来扩展

• 与 Obsidian 的双向链接功能集成，自动生成知识图谱。
• 每日自动生成 memory/YYYY-MM-DD.md 的摘要。
• 支持导出 memory.db 为 CSV 或 JSON 以供备份。
• 使用更先进的 NLP 模型（如本地小模型）来增强 query_memory.py 的语义理解能力。

“记忆不是用来记住的，是用来被访问的。” —— 你的数字大脑