可信 — 风险评分 0/100
上次扫描:1 天前 重新扫描
0 /100
memory-vault
Privacy-first persistent memory for AI agents with 3 privacy tiers (Open/Local/Locked), hierarchical recall, and keyword + semantic search
Memory Vault is a documentation-only skill with no executable code. It describes a legitimate privacy-focused persistent memory system with no malicious indicators detected.
技能名称memory-vault
分析耗时24.3s
引擎pi
可以安装
This skill is safe to use. However, it requires additional implementation files (Python modules) to be functional. Verify the actual implementation from the GitHub repository before deployment.
资源类型声明权限推断权限状态证据
文件系统 NONE NONE SKILL.md: No file operations in documentation
网络访问 READ NONE ✓ 一致 SKILL.md: External URLs are informational (ollama.ai, vivioo.io)
命令执行 NONE NONE SKILL.md: No shell commands documented
环境变量 NONE NONE SKILL.md: No environment variable access
技能调用 NONE NONE SKILL.md: No cross-skill invocation
剪贴板 NONE NONE SKILL.md: No clipboard access
浏览器 NONE NONE SKILL.md: No browser automation
数据库 NONE NONE SKILL.md: ChromaDB would be used for vectors, but no code present
2 项发现
🔗
中危 外部 URL 外部 URL
https://vivioo.io
SKILL.md:12
🔗
中危 外部 URL 外部 URL
https://ollama.ai
SKILL.md:58

目录结构

1 文件 · 15.7 KB · 465 行
Markdown 1f · 465L
└─ 📝 SKILL.md Markdown 465L · 15.7 KB

安全亮点

✓ Documentation-only skill with no executable code present
✓ Declares legitimate dependencies (chromadb, cryptography) in metadata
✓ Describes a privacy-first architecture with 3-tier isolation (Open/Local/Locked)
✓ Local-tier data explicitly designed to never reach LLM context
✓ Locked-tier uses Fernet encryption for sensitive data
✓ No obfuscation, encoding tricks, or suspicious patterns detected
✓ No credential harvesting mechanisms
✓ No data exfiltration pathways identified
✓ GitHub and homepage URLs are legitimate services