Thanks — good question. You're asking whether it's safe to

await session.say(..., allow_interruptions=False)

from inside a tool call. Yes — you can safely await

session.say(...)

from within a tool.

session.say()

returns a

SpeechHandle

that you can await to wait until the spoken audio finishes playing. Using

allow_interruptions=False

on that call ensures the particular utterance is not interrupted by user input while it playouts. If your tool is long-running and you want to prevent any user speech from interrupting any part of the tool (not just that single

say

), you can also call

run_ctx.disallow_interruptions()

at the start of the tool to enforce no interruptions for the tool's duration. Using both approaches together keeps behavior consistent: the

SpeechHandle

await lets your tool continue only after speech completes, and

run_ctx.disallow_interruptions()

prevents user input from interrupting the broader tool execution. If you want, tell me: • which SDK/language you're using, and • whether the tool needs to remain responsive to other events during the wait and I can provide a short code example or point out any subtle concurrency considerations for that environment. Sources: Agent speech and audio | LiveKit Documentation | Uninterruptable Agent | LiveKit Documentation | Tool definition and use | LiveKit Documentation