ワンクリックで
service-management
Use when inspecting, starting, stopping, restarting, or reading logs for managed Linux/macOS services through SysGuard.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
メニュー
Use when inspecting, starting, stopping, restarting, or reading logs for managed Linux/macOS services through SysGuard.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Use when creating structured incident alerts from SysGuard findings, health reports, log analysis, remediation failures, or operational events.
Use when checking database connectivity, running read-only diagnostic SQL, validating DSNs, or collecting small database health evidence through SysGuard.
Use when reading, statting, listing, or tailing local files for SysGuard diagnostics without modifying filesystem contents.
Use when checking host health, service health, CPU, memory, disk, network status, or determining whether SysGuard should treat a system as unhealthy.
Use when analyzing operational logs, filtering errors or warnings, summarizing incidents from log files, or finding relevant lines in SysGuard-managed logs.
Use when collecting structured CPU, memory, disk, network, service, or health-score metrics from SysGuard for dashboards or reports.
| name | service-management |
| description | Use when inspecting, starting, stopping, restarting, or reading logs for managed Linux/macOS services through SysGuard. |
Use this skill for explicit service operations after health checks or log analysis identify a service-level issue.
The Go implementation is registered by RegisterCoreSkills and invoked as "service-management":
registry.Execute(ctx, "service-management", &skills.SkillInput{Params: map[string]interface{}{
"operation": "status",
"service": "nginx",
"lines": 100,
}})
Supported operations are status, logs, start, stop, and restart. Non-Linux systems currently support status via process lookup.
status or logs before changing service state.stop, start, or restart, explain the intended action and approval requirement.allow_dangerous=true only after explicit user/operator approval.status after a state-changing operation.State-changing commands may be blocked by the runtime CommandInterceptor. Never bypass dangerous-command checks by constructing arbitrary shell commands.