Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

Service Persistence

v1.1.0

macOS 服務持久化與重啟恢復。管理 LaunchAgent、tmux bootstrap、wrapper daemon 三層架構, 確保電腦重啟後所有服務自動恢復。Use when: (1) 新增需要常駐的服務,(2) 把現有 tmux session 升級成 LaunchAgent, (3) 為互動式程式建...

0· 61·0 current·0 all-time
byJoey Luo@darwin7381
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Pending
View report →
OpenClawOpenClaw
Suspicious
high confidence
!
Purpose & Capability
The skill's core goal (managing LaunchAgents, tmux bootstrap, wrapper daemons) aligns with required actions (copy plists, create scripts, run launchctl). However included artifacts are highly personalized (hard-coded /Users/btai HOME and paths) and one wrapper specifically launches a Claude process with flags '--dangerously-skip-permissions --permission-mode bypassPermissions', which goes beyond 'service persistence' and introduces elevated access to user resources not justified by a generic persistence tool.
!
Instruction Scope
SKILL.md instructs copying scripts into ~/.openclaw, deploying plists to ~/Library/LaunchAgents, and running launchctl — all expected. But wrapper logic will capture tmux panes and automatically send Enter for 'trust this folder' prompts and automatically restart services; auto-confirming trust prompts and running an LLM with bypass flags gives the skill direct capability to approve prompts or re-start privileged processes without manual oversight.
Install Mechanism
No remote install or downloads are declared (instruction-only skill with embedded scripts). That lowers supply-chain risk: nothing fetched from arbitrary URLs. The install actions described (copying files to ~/Library/LaunchAgents and ~/.openclaw) are local and expected for this purpose.
!
Credentials
The skill declares no required env vars, but the included scripts hardcode HOME and PATH to a specific user (/Users/btai) and add user-local binary directories (e.g., ~/.bun/bin). Hardcoding a user account and adding user paths is not proportional for a generic skill and may cause the skill to run unintended binaries; the claude invocation with permission-bypass flags is especially disproportionate for a persistence helper.
Persistence & Privilege
The skill does not set always: true, but it instructs installing LaunchAgents (RunAtLoad / KeepAlive) which will give long-lived, autonomous execution under the installing user account. That is consistent with its purpose but is powerful: any wrapper/daemon placed under ~/Library/LaunchAgents will automatically run and persist across reboots.
What to consider before installing
What to consider before installing: - Inspect and edit the scripts before deploying: replace hard-coded HOME (/Users/btai) and any absolute paths with variables or your own paths so they don't act on someone else's environment. - Pay special attention to scripts/restore-claude-telegram.sh: it launches a Claude process with '--dangerously-skip-permissions --permission-mode bypassPermissions' and a Telegram plugin. These flags likely disable built-in permission checks and could allow the process to access files, network endpoints, or credentials without further prompts. Remove or understand these flags before use. - The wrapper auto-accepts 'trust this folder' prompts by sending Enter to tmux panes. This can silently approve prompts that affect file access or authorization—remove auto-confirm behavior or require manual intervention for sensitive prompts. - Test in an isolated account or VM first. Verify what each LaunchAgent will run, check logs in ~/.openclaw/logs, and confirm no unexpected network connections or access to secrets occur. - If you only need non-interactive services, prefer Tier 1 LaunchAgents with explicit ProgramArguments and avoid tmux wrappers. For interactive services, require manual approval rather than automatic trust/confirmation. - After deployment, review ~/Library/LaunchAgents entries and the actual plist contents; uninstall (launchctl unload and remove plist) if behavior is unexpected. If you want, I can (1) produce a sanitized version of these scripts that remove auto-confirm and permission-bypass flags, (2) generate a checklist of exact lines to change for your username and paths, or (3) walk through a safe test plan to validate the skill in a sandbox.

Like a lobster shell, security has layers — review code before you run it.

latestvk97egtk35zmm80a9bmqpeyywe184h5d2

License

MIT-0
Free to use, modify, and redistribute. No attribution required.

Comments