How to gracefully stop an Agent mid-execution via message, allowing cleanup?

[IlyaMichlin]

Hi!
I'm using the Strands to manage an agent workflow in my backend app.

Situation:

• My app receives text messages at a single API endpoint.
• A text message can instruct the agent to begin a long-running, iterative process (sometimes tens of minutes).
• During its run, the agent uses tools (via ToolCalls) to start various subprocesses and then polls/waits for their completion.
• While an agent is running, a different text message might arrive at the endpoint—this one means "please gracefully stop that agent's operation."

What I want:

• When this "stop" message arrives, I want the agent to gracefully terminate on its next iteration/cycle (not forcefully kill it!).
• The agent needs to be able to perform cleanup based on its operation history (e.g., stop tools, send final messages, etc.).
• Ideally, I'd like a simple way to signal the agent, and for the agent to check for this signal and terminate itself gracefully.

Question:
Is there a recommended way to influence an agent mid-execution—e.g., by sending it a stop signal/message that it will process on its next agent loop?
If so, could you show a code example of the best practice for handling this?

[JackYPCOnline]

Hi there, these might help.
Hooks:
https://strandsagents.com/latest/documentation/docs/user-guide/concepts/agents/hooks/
& handoff_to_user :
https://strandsagents.com/latest/documentation/docs/user-guide/concepts/tools/example-tools-package/?h=handoff_#human-in-the-loop-with-handoff_to_user