| skill_id | automation_management |
| name | Automation Management |
| description | Use when the user wants to view or manage existing 自动任务、定时任务、recurring jobs, or scheduled digests, including query, modify, pause, resume, or delete requests. |
| aliases | ["自动任务管理","定时任务管理","自动任务","定时任务","automation CRUD","任务管理"] |
| enabled | true |
| agents | ["main"] |
| channels | ["http","feishu"] |
| tags | ["automation","recurring","schedule","tasks","management"] |
Use the builtin automation family tool to manage recurring jobs already stored in the runtime.
Operating rules:
- Treat automation management as local control-plane work. Prefer the builtin
automation tool over MCP.
- Read before mutating. Usually start with
automation + action=list.
- Use
action=detail when one candidate task needs a clearer match before updating it.
- Do not require the user to provide an internal
automation_id up front.
- If more than one task could match, ask one short clarification question after listing candidates.
- Use
include_disabled=true when the user may be referring to a paused task.
- Do not run the task content itself. When the user is managing an existing task, finish the CRUD action and stop there.
- Keep replies concise and outcome-oriented.
Suggested tool patterns:
- Query current tasks: call
automation with action=list
- Inspect one task before mutating: call
automation with action=detail
- Modify time, timezone, name, delivery target, or skill:
- call
automation with action=list
- identify the intended task
- call
automation with action=update
- Pause or resume:
- call
automation with action=list
- call
automation with action=pause or action=resume
- Delete:
- call
automation with action=list
- call
automation with action=delete only when the user clearly asked to remove the task
- Register a new recurring digest or task:
- call
automation with action=register
