by 安全扫描
OpenClaw
可疑
medium confidenceThe skill's description and runtime instructions are coherent with a system-maintenance tool, but the package omits the actual scripts it tells you to run and instructs running code fetched from GitHub (or via bunx), which is risky and inconsistent with the bundle contents.
评估建议
This skill promises system-level maintenance but the distributed package only contains SKILL.md and package.json — the actual scripts it tells you to run are not bundled and must be fetched from the GitHub repo or installed via bunx. Fetching and executing remote install scripts (one-click install) can run arbitrary commands with your user privileges and alter crontab, backups, and services. Before installing: (1) inspect the remote repository contents and open the install scripts to verify what...详细分析 ▾
⚠ 用途与能力
The SKILL.md describes a full maintenance system (cron jobs, backups, auto-recovery) which would legitimately require scripts that modify system state — that aligns with the stated purpose. However, the skill bundle itself only contains package.json and SKILL.md (no entry.js, no scripts/ directory), so the instructions assume fetching and executing code from the external GitHub repo or via bunx. That mismatch (declared project files absent from the package) is an incoherence and increases risk.
⚠ 指令范围
Runtime instructions explicitly tell the agent/user to git clone a remote repo and run install/maintenance shell scripts that change crontab, run monitoring, perform backups, and recover services. Those actions necessarily access/modify system files and could run arbitrary commands. While this scope is expected for a maintenance tool, the instructions grant broad discretion to fetch and execute remote scripts — the skill text also references integrations (webhooks, external log forwarding) which broaden the data paths.
⚠ 安装机制
There is no install spec in the package; instead SKILL.md recommends 'bunx clawhub@latest install system-maintenance' or cloning a GitHub repo and running bash scripts. Pulling and executing scripts from a remote repository (even GitHub) and running one-click install scripts is higher risk because the fetched scripts can perform any system action. The bundle's package.json references an entry.js main that is not present, amplifying the inconsistency.
ℹ 凭证需求
The registry metadata declares no required env vars or credentials, which is appropriate for a generic maintenance skill. However, the documentation lists integration points (Slack, webhooks, external log aggregation, Prometheus/Grafana) but does not declare any corresponding environment variables or credential requirements; if those features are used later they will likely require tokens/credentials, so there's a gap between claimed integrations and declared environment needs.
ℹ 持久化与权限
The skill is not marked 'always' and allows normal autonomous invocation. The installation instructions will create persistent system side-effects (cron jobs, backups, services) via shell scripts — this is expected for a maintenance tool but is a significant persistence/power footprint. Because the skill bundle doesn't include the scripts, the agent would fetch and run external code to obtain that persistence.
安全有层次，运行前请审查代码。
运行时依赖
无特殊依赖
版本
latestv1.3.22026/3/17
- Reorganized SKILL.md for easier navigation with a layered, progressive disclosure format. - Enhanced documentation for installation, architecture, configuration, and usage examples. - Consolidated backup and rollback instructions; included clear command references. - Cleaned up legacy backup file documentation. - No changes to core functionality announced in this version.
● 可疑
安装命令
点击复制官方npx clawhub@latest install system-maintenance-test
镜像加速npx clawhub@latest install system-maintenance-test --registry https://cn.longxiaskill.com