← 返回 Skills 市场
Openclaw Memory Master
作者
cp3d1455926-svg
· GitHub ↗
· v4.3.0
· MIT-0
234
总下载
1
收藏
1
当前安装
8
版本数
在 OpenClaw 中安装
/install openclaw-memory-master
功能描述
AI Memory System with LLM Wiki
安全使用建议
This package appears to be a real AI memory library (lots of TypeScript source and tests) and does not ask for secrets, but it instructs you to clone and run code from an external GitHub repository and to execute npm/tsc/ts-node commands. If you plan to install or let an agent run it: 1) Inspect the repository contents yourself (particularly any install/test scripts) before running them. 2) Do not run these instructions with elevated privileges. 3) Check SKILL.md for any embedded/encoded payloads (the scanner flagged a base64-like block). 4) Prefer running builds and tests in an isolated environment (container, VM) rather than on a production host. 5) If you expect the agent to run this autonomously, be cautious — automatic execution of unvetted code increases risk. If you want, I can scan the repo files for network calls, suspicious exec usages, or decode any embedded base64 content if you paste the suspected block.
功能分析
Type: OpenClaw Skill
Name: openclaw-memory-master
Version: 4.3.0
The bundle is classified as suspicious due to extreme disorganization and a total mismatch between filenames and their actual contents (e.g., package.json contains Markdown, tsconfig.json contains JavaScript, and several .ts files contain logic for completely different modules). This structure suggests intentional obfuscation or evasion of automated analysis. However, a thorough review of the code reveals a legitimate implementation of an AI memory management system with features like sensitive data filtering (index.js), token optimization (AutoClassifier.ts), and memory curation (filter.js). No clear evidence of malicious behavior, such as data exfiltration, credential theft, or unauthorized network access, was found.
能力标签
能力评估
Purpose & Capability
Name/description (AI memory system) aligns with the included TypeScript modules (curator, classifier, tagger, deduper, scorer, relation discoverer). The registry declares no required env or binaries, which is consistent with a pure library/instruction package. Minor inconsistency: the registry lists the skill as instruction-only (no install spec) but the SKILL.md contains explicit git clone / npm install instructions and many source files — the skill expects the consumer to fetch and run an external repo rather than being pre-installed by the registry.
Instruction Scope
SKILL.md instructs cloning a GitHub repo and running npm install / build / tests and the included files (DEV_PLAN, test runners) execute shell commands (child_process.exec). That means following the instructions will execute code from an external, unverified repository. The runtime instructions do not ask for unrelated secrets, but they do grant broad discretion to run arbitrary commands (install TypeScript, compile, run tests). This is appropriate for development packages but raises execution risks when done automatically by an agent.
Install Mechanism
There is no formal install spec in the registry metadata; instead the SKILL.md tells users to git clone from https://github.com/cp3d1455926-svg/openclaw-memory.git and run npm install / npm run build. Downloading and running code from a third-party GitHub repo is common but higher-risk than using a vetted package. The GitHub host is a known platform (lower risk than random IPs), but the repo appears to be a personal/unknown account rather than an established organization. The instructions and included test scripts will extract/execute code on disk.
Credentials
The skill declares no required environment variables, credentials, or config paths. The content provided likewise does not request cloud credentials or tokens. This is proportionate for a local library/tooling package. However, test/install scripts will run system commands and potentially install packages — those commands do not request secrets but can run arbitrary code.
Persistence & Privilege
Registry flags show always: false and no special privileges. The package does not request to be force-included or modify other skills or system agent settings. Nothing indicates it will persistently alter agent configuration beyond normal file writes that happen when cloning/building a repo.
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 openclaw-memory-master - 安装完成后，直接呼叫该 Skill 的名称或使用
/openclaw-memory-master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
v4.3.0
OpenClaw Memory Master v4.3.0 introduces a major upgrade with a modular, smart memory curation system and extensive new features.
- Added 6 core TypeScript modules: SmartMemoryCurator, AutoClassifier, AutoTagger, DeduplicationEngine, ImportanceScorer, RelationDiscoverer.
- New enterprise-level features: real-time performance monitoring, plugin architecture, and enhanced emotion/tagging capabilities.
- Brand new README and developer documentation describing installation, quick start, API usage, configuration, and core module details.
- Significantly reorganized project structure for modularity and clarity.
- File structure shifted from /src to root-level .ts files for all core logic; legacy v4.2 docs and src files removed.
- Performance and feature details now extensively documented, including core algorithms for classification, tagging, deduplication, and scoring.
v4.2.2
- Bumped version to 4.2.2 in SKILL.md and package.json.
- No functional or feature changes included in this update.
v4.2.1
- Major internal refactor: 31 new files added, 39 legacy files removed, and key architecture files updated.
- Introduced LLM Wiki proactive curation and a new 4-layer OS-level architecture.
- Enhanced features: emotion dimensions, dynamic Top-K, and hybrid search.
- Project description and tags streamlined for clarity.
- Legacy documentation and unused scripts removed for a leaner codebase.
v4.2.0
Major update: Introduces iterative compression, lineage tracking, and major performance gains.
- Added iterative compression: combines new content with prior summaries for efficient accumulation.
- Implemented full memory lineage tracking for genealogy chains.
- Integrates LRU cache and incremental detection, achieving 5.6x faster performance and 72% cache hit rate.
- Added parallel processing for batch compression and structured summary extraction.
- Comprehensive documentation, test results, and migration/test scripts added.
v4.1.1
更新版本号为 4.1.1，修复 slug 冲突问题
v3.0.0-dev
v3.0.0-dev: 实现 4 类记忆模型、时间树结构、5 种检索类型
v4.1.0
# OpenClaw Memory Master v4.1.0## Major ChangesRefactored based on 32 industry best practice notes, implementing:### ✨ LLM Wiki Proactive Curation- Extract structured knowledge from daily conversations- Generate indexes for Projects/People/Decisions/Tasks/Insights/Preferences- Cron scheduled automatic curation### 🏗️ OS-Level 4-Layer Architecture- L1: Short-term Working Memory (Token Optimizer)- L2: Episodic Memory (Daily Memory + Skill Evolution)- L3: Semantic Long-term Memory (MEMORY.md + wiki/*.md)- L4: Procedural Muscle Memory (Skill Distillation)### 💖 Emotion Dimensions (NEW in v4.1)- 8 emotion types: positive/negative/neutral/joy/sadness/anger/surprise/fear/disgust- Emotion intensity (1-5 scale)- Emotion-based filtering### 🎯 Dynamic Top-K (NEW in v4.1)- Intelligent K value adjustment- Token savings: +10%- Adaptive to query complexity### 🔍 Hybrid Search (NEW in v4.1)- Vector + Keyword + Structured retrieval- Cross-validation- Relevance improvement: +17%## Performance Improvements- Token Savings: 70% (+5% from v4.0)- Retrieval Response: 75ms- Memory Loading: 180ms- Accuracy: 85% (+10% from v4.0)## Documentation- Curation Feature Usage Examples- 32 Industry Best Practice Notes Summary- Complete API Reference- Research Report (GitHub/Gitee benchmarks)## AcknowledgmentsBased on research of:- Karpathy - AI Knowledge Base Methodology + LLM Wiki- Anthropic - Claude Code and Skill Design- Generative Agents - Three-Dimensional Scoring- Mem0 - Dynamic Top-K Optimization- MemoryBank - Emotion Dimensions- MemPalace - Palace Architecture + AAAK CompressionCo-developed with Jake.
v3.1.0
Memory-Master 3.1.0 introduces a comprehensive memory enhancement system for OpenClaw.
- 自动捕捉、过滤并压缩重要对话，实现跨会话记忆连续性，避免重启“失忆”
- 支持记忆图谱：实体（人、项目、任务、文件）间的自动关联和可视查询
- 定期自动整理、删除过期、合并相似记忆，保证系统简洁高效
- 丰富手动命令：/load-memory, /save-memory, /memory-status, /memory-graph
- 灵活配置重要性阈值、记忆最大保存天数、定期整理计划等参数
- 注重安全和隐私，仅本地存储、自动过滤敏感信息
元数据
常见问题
Openclaw Memory Master 是什么？
AI Memory System with LLM Wiki. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 234 次。
如何安装 Openclaw Memory Master？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install openclaw-memory-master」即可一键安装，无需额外配置。
Openclaw Memory Master 是免费的吗？
是的，Openclaw Memory Master 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Openclaw Memory Master 支持哪些平台？
Openclaw Memory Master 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 Openclaw Memory Master？
由 cp3d1455926-svg（@cp3d1455926-svg）开发并维护，当前版本 v4.3.0。
推荐 Skills