记忆秘书

智能管理与优化记忆文件，提供质量检查、重复任务检测、成功案例提取、工作模式分析和智能提醒。

This package is internally inconsistent: docs/tests reference many files and scripts that are not included, and the code—contrary to SKILL.md—creates and writes files under your workspace/memory. The README also tells users to force-install past VirusTotal warnings. Before installing, do the following: (1) inspect the full source yourself and confirm which files will be created under your workspace/memory and what they contain; (2) run the code in an isolated sandbox or VM with a backed-up test workspace (not your real memory data); (3) verify missing modules/scripts (smart_adaptive, daily_check, pilot_check, scripts/) — ask the author for a complete release or updated manifest; (4) do not use --force to bypass security warnings without code review; (5) if you proceed, point workspace_root to a dedicated test directory and check precisely what files are written. These contradictions are suspicious but do not prove intent to harm; require clarification or a corrected release.

Type: OpenClaw Skill Name: memory-secretary-v1-0-1 Version: 1.0.1 The skill bundle is classified as suspicious primarily due to instructions in README.md and CHANGELOG.md that explicitly advise users to ignore security warnings (VirusTotal) and use installation overrides (--force). While the core logic in src/memory_secretary_lite.py appears to be a benign tool for analyzing local markdown files using Python standard libraries, the use of social engineering to dismiss security alerts is a classic indicator of potentially harmful intent or hidden payloads. No direct evidence of data exfiltration or malicious execution was found in the provided files, but the attempt to undermine security tooling warrants caution.