PM Agent not automatically reviewing its own created issues or recovering from failed tasks

[fenfenxu]

I have a PM agent in Multica whose main role is to create issues and assign them to worker agents. The PM agent is automatically added as a subscriber to all issues it creates. However, it does not automatically intervene when issue status changes, which is not matching my expectation.
The relevant part of my agent definition is:

## subscribe issue
subscribe 的 issue，如果发生了状态变更，自主判断是否要介入协助处理。

- 比如处于待评审状态，如果是自己创建的那么就自己 review，并推进后续诸如包括但不限于给当前执行 agent反馈review的问题，merge pr，close issue ，升级给人类介入（刘旭）。
- 比如发生了执行失败 或中断，就尝试诊断协助解决恢复 issue的执行，包括但不限于retry、re-assign、升级给人类介入（xx）。

Expected behavior:

1. When an issue created by this PM agent enters pending review, the PM should automatically review it, give feedback to the worker agent, merge PR, close issue, or escalate to a human.
2. When an issue fails or is interrupted, the PM should diagnose, recover, retry, reassign, or escalate.

Current situation:

• The PM agent is listed as a subscriber on all issues it creates.
• No automatic review happens for pending review status.
• No failure recovery is triggered when executions fail.
• I can make this work with autopilot, but it’s based on scheduled polling (cron), which is not real-time and causes delays.

Question / Feature request

Does Multica currently support a status change hook / event-driven trigger for agents? That is, can an agent be set to immediately wake up and act when a subscribed issue changes status, instead of relying on periodic autopilot?

If this is not supported today:

• Is there any existing configuration or workaround to achieve real-time reaction?
• Are there plans to add a status change hook or subscription-based event trigger in the roadmap?

Thanks