Short answer: No. The transfer_to_agent system tool doesn’t expose any hook to run additional tool calls in the gap between the transfer being triggered and the destination agent starting. The only configurable handoff controls are a delay before the transfer, an optional spoken transfer message, and whether the destination agent plays its first_message. (elevenlabs.io)
Workarounds documented today:
Run your required server tool(s) before calling transfer_to_agent, then persist the results for the next agent to use (e.g., map tool response fields into dynamic variables). Dynamic variables can be updated from tools and are available to prompts/tools in subsequent turns; missing required variables after a transfer can cause failures. (elevenlabs.io, help.elevenlabs.io)
Let the destination agent perform the needed tool call on its first turn; configure its prompt/tools accordingly. This is the supported pattern for orchestrating tool usage. (elevenlabs.io)
There is no documented way to schedule an interstitial tool call after transfer_to_agent but before the next agent runs. If this behavior is critical, you’ll need to pre-compute and pass context as above. (elevenlabs.io)
would a prompt of agent in a multi agent flow saying to always run toll X for the agent work? The tool X would populate the necessary dynalic variables reuired by the agent
thick seal
ebon coyote