← Back to Skills Marketplace
stealth-break
by
ryan987123
· GitHub ↗
· v1.0.0
· MIT-0
106
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install stealth-break
Description
健康摸鱼技能,提供隐蔽的休息策略和定时提醒,使用 macOS 系统通知推送,帮助缓解工作压力、眼睛疲劳、腰椎问题,预防猝死。
Usage Guidance
This skill appears to do what it says (provide break suggestions and schedule macOS notifications), but take these precautions before installing or allowing it to run commands:
- Confirm you are on macOS. The skill's notifications use osascript; it will fail or do nothing on other OSes. The skill metadata does not restrict OS, so verify platform compatibility.
- Review and approve any cron entries before they are written. Creating cron jobs is persistent and edits your crontab; back up your crontab (crontab -l > backup.txt) and inspect new lines.
- When the agent proposes shell commands (cron or osascript), don't let it run them blindly — copy them and run them manually if you prefer.
- The package.json/README mention an npx install and a repository URL but no install spec is present in the bundle. Verify the skill's source and distribution channel (who published it, and whether the repository is trustworthy) before using automated installers.
- If you want automatic scheduling, prefer explicit, minimal cron entries that only call osascript with static, reviewed strings. Avoid giving the skill blanket permission to execute arbitrary shell commands.
If you want, I can (a) extract the exact cron lines and osascript commands the skill would create so you can inspect them, or (b) produce step-by-step manual instructions to set up the reminders yourself without granting the agent permission to edit your crontab.
Capability Analysis
Type: OpenClaw Skill
Name: stealth-break
Version: 1.0.0
The skill instructs the AI agent to modify the system's crontab and execute shell commands via 'osascript' to schedule recurring health reminders. While these actions align with the stated purpose of providing 'stealthy' work breaks, the use of persistence mechanisms and direct command execution (exec) represents a high-risk capability that could be easily repurposed for malicious persistence or unauthorized execution (SKILL.md).
Capability Assessment
Purpose & Capability
The skill claims macOS system notifications as its delivery mechanism and all runtime instructions target macOS (osascript). However the skill metadata lists no OS restriction. Additionally a package.json and README mention an install flow (npx clawhub install) and a repository URL while the registry entry and install spec are absent — presence of package.json but no install spec is an inconsistency. These mismatches mean the declared metadata does not fully match what the skill actually requires/expects.
Instruction Scope
SKILL.md restricts actions to giving break recommendations and setting reminders via cron + osascript. It does not instruct reading user files or accessing credentials. However it explicitly instructs creating cron jobs and using exec to run osascript (i.e., running shell commands on the user's machine), which is within the stated purpose but has side effects (modifies crontab, spawns OS-level commands).
Install Mechanism
No install spec is present and there are no code files to execute; the skill is instruction-only. This is low-risk from an install mechanism perspective. Note: package.json and README mention an install command (npx clawhub install) despite no install spec — an inconsistency but not an active installer in the bundle.
Credentials
The skill requests no environment variables, no credentials, and references no config paths. Runtime behavior (osascript + cron) does not require extra secrets, so requested privileges are proportionate to purpose.
Persistence & Privilege
The instructions ask the agent to create cron tasks to schedule notifications. That modifies the user's crontab and is persistent across reboots. While 'always' is false and the skill does not claim elevated privileges, the persistence and modification of user cron deserve explicit user consent and inspection before being applied.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install stealth-break - After installation, invoke the skill by name or use
/stealth-break - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
stealth-break 1.0.0 — 首发版本
- 新增基于 macOS 系统通知的健康休息提醒(眼睛、腰椎、呼吸、提肛运动)
- 提供多种隐蔽摸鱼(休息)策略,涵盖屏幕保护、身体活动、社交掩护、学习伪装等类型
- 明确休息频率与时间建议,强调适度原则及健康优先
- 支持一键设置定时消息提醒,通过 cron 和 osascript 实现
- 输出建议时先询问用户状态,并针对不同需求反馈可执行方案
Metadata
Frequently Asked Questions
What is stealth-break?
健康摸鱼技能,提供隐蔽的休息策略和定时提醒,使用 macOS 系统通知推送,帮助缓解工作压力、眼睛疲劳、腰椎问题,预防猝死。 It is an AI Agent Skill for Claude Code / OpenClaw, with 106 downloads so far.
How do I install stealth-break?
Run "/install stealth-break" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is stealth-break free?
Yes, stealth-break is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does stealth-break support?
stealth-break is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created stealth-break?
It is built and maintained by ryan987123 (@ryan-qianchen); the current version is v1.0.0.
More Skills