deterministic / consistent first sentence for subagents

### Before filing - [x] I've read `SCOPE.md` and checked `.out-of-scope/`, and this isn't covered by either. - [x] This issue is about one problem. - [x] I searched open and closed issues for a duplicate. ### Skill every agent triggerable skills ### What went wrong Generally 2/10 ticket implementor agents goes upto 400k context. Okay, it may look like about ticket sperating problem. But actually, the agent and even the developer, may not know the ticket is not big or small. So, this is normal. But it brings large context side effects to us. (opus 5.5 / claude code cli) ### Proposed change Can the sub-agents spawned by your skills start their first sentence with [[agent-type]] or You are a [[agent-type]]...? This would allow people to write hooks that can detect what type of sub-agent is running through a simple string search, if they want to. What they do with this information is up to their imagination. I am absolutely NOT asking you to introduce a new hook. Consistent sub-agent openings or a marker somewhere in their output that I can catch with a regex would be more than enough. For example, I’m planning to solve the context overload problem by writing a hook where I can specify which agent types to detect, then have the hook instruct the orchestrator to make those agents finish with a handoff and restart them.