← 返回 Skills 市场
131
总下载
1
收藏
0
当前安装
3
版本数
在 OpenClaw 中安装
/install six-layer-memory
功能描述
Set up or repair a proactive six-layer memory system for an OpenClaw/Codex workspace. Use when a user wants durable HOT/WARM/COLD/CURATED/CLOUD/AUTO memory,...
安全使用建议
This skill appears to implement what it claims. Important things to consider before installing: 1) Cloud sync: if you place a SuperMemory key in memory/auto-extract/.supermemory_key the supermemory_sync script will upload workspace memory files (SESSION-STATE.md, MEMORY.md, daily logs) to https://api.supermemory.ai — only add that key if you trust the external service and you want those files uploaded. 2) AUTO: if you have a ~/.mem0/config.json and the mem0 Python client is available, the scripts will read that config and may send extracted facts to your Mem0 backend; review ~/.mem0/config.json and the mem0 client to ensure it’s configured as you expect. 3) Cron/autonomy: the README suggests cron entries that run every 30 minutes; installing those will cause regular automatic reads and (if configured) uploads — audit the cron jobs you create. 4) Review workspace contents: the scripts will execute other helper scripts in the workspace (e.g., lancedb_sync or transcript import) via subprocess; ensure those helper scripts are trusted. 5) File placement and permissions: the installer sets SESSION-STATE.md and MEMORY.md to 600/644; verify file permissions if multiple users share the host. If you want to proceed, review/modify scripts (especially supermemory_sync.py and mem0_sync.py) to match your privacy and operational requirements, and avoid placing sensitive secrets inside workspace memory files.
功能分析
Type: OpenClaw Skill
Name: six-layer-memory
Version: 2.0.1
The 'six-layer-memory' skill is a legitimate tool for managing agent memory across multiple layers, including local files and optional cloud sync. It includes scripts for setup (install_workspace.sh), maintenance (auto_memory_6layer.py), and syncing with external services like SuperMemory (supermemory_sync.py) and Mem0 (mem0_sync.py). All high-risk behaviors, such as external API calls to api.supermemory.ai and cron-based automation, are clearly documented as core features and require explicit user configuration, such as providing an API key in .supermemory_key. The scripts follow reasonable security practices, such as applying restrictive file permissions (chmod 600) to sensitive memory files.
能力标签
能力评估
Purpose & Capability
Name/description (six-layer memory, HOT/WARM/COLD/CURATED/CLOUD/AUTO) match the bundled scripts and installer: installer creates memory files, maintenance runner scans session state, extracts facts, syncs to Mem0, and can upload to SuperMemory. No unrelated binaries or credentials are requested.
Instruction Scope
SKILL.md instructs installation of scripts, cron scheduling, and optionally placing a SuperMemory key in memory/auto-extract/.supermemory_key. The runtime scripts read workspace files (SESSION-STATE.md, MEMORY.md, daily logs), call other helper scripts in the workspace, and may read the user home mem0 config (~/.mem0/config.json) for AUTO. This behavior is expected for AUTO/CLOUD features but means the skill will read user memory files and (if configured) upload them externally.
Install Mechanism
No install spec (instruction-only) — installer is a simple shell script included in the bundle that copies the scripts into the target workspace. Nothing is downloaded from external URLs and no archive extraction or third-party package installs are performed by the skill itself.
Credentials
The skill declares no required env vars or primary credential. At runtime it will optionally read ~/.mem0/config.json (AUTO) and will attempt to read a SuperMemory key file under workspace memory/auto-extract/.supermemory_key (CLOUD) if present. Those accesses are proportional to the AUTO and CLOUD features but are not declared as required in metadata — users should be aware the mem0 config and a workspace key file are read if present.
Persistence & Privilege
Flags: always=false, user-invocable=true, agent invocation allowed. The skill does not request permanent inclusion or modify other skills. Cron instructions create periodic execution but that is user-managed; the skill itself does not force persistent platform privileges.
如何使用
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install six-layer-memory - 安装完成后,直接呼叫该 Skill 的名称或使用
/six-layer-memory触发 - 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v2.0.1
Improved reliability and publishability for proactive six-layer memory. This release keeps the same six-layer model but tightens installation, packaging, bilingual beginner guidance, and multi-workspace maintenance so users can adopt it with less manual cleanup.
v2.0.0
Rebuilt six-layer-memory as a proactive per-workspace system with unified HOT/WARM/COLD/CURATED/CLOUD/AUTO maintenance, optional LM Studio or Ollama AUTO backends, generic SuperMemory sync, and workspace-safe flush support.
v1.0.0
Six-Layer Memory System Skill 1.0.0
- Introduces a hierarchical six-layer memory architecture inspired by human cognition, with HOT, WARM, COLD, CURATED, CLOUD, and AUTO layers working collaboratively for AI memory management.
- Implements WAL (Write-Ahead Logging) protocol: user states and decisions are committed before responses are generated.
- Details the responsibilities and triggering conditions for each layer to manage active session state, semantic search, structured decisions, curated logs, cross-device sync, and real-time fact extraction.
- Provides structured directory guidance and core script descriptions for seamless agent integration.
- Offers usage guidelines for main agents and knowledge bots, and explains compatibility with other memory-related skills.
元数据
常见问题
Six-Layer Memory 是什么?
Set up or repair a proactive six-layer memory system for an OpenClaw/Codex workspace. Use when a user wants durable HOT/WARM/COLD/CURATED/CLOUD/AUTO memory,... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 131 次。
如何安装 Six-Layer Memory?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install six-layer-memory」即可一键安装,无需额外配置。
Six-Layer Memory 是免费的吗?
是的,Six-Layer Memory 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
Six-Layer Memory 支持哪些平台?
Six-Layer Memory 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。
谁开发了 Six-Layer Memory?
由 曹操(@ashu2025-rgb)开发并维护,当前版本 v2.0.1。
推荐 Skills