You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Instead of having human input only via agent making a tool call and the user input in the response, instruct the agents that user input may come at any time in the thread, but they must use a tool to specifically request a response from the user.
The pattern the agent would see is:
...
<agent used human input tool: 'hello, whats your name?'>
<tool response successful>
<human message>
The key is we have to show the tool call so that agents don't conclude from history that they can converse with users by posting "regular" messages.
This will also facilitate adding user messages at any point, perhaps with a mechanism like
agent.send_message()
or
task.send_message()
sending a global message doesn't seem necessary at the moment.
The text was updated successfully, but these errors were encountered:
yes please! I hadn't considered the "create a synthetic tool call before inserting the user input" to allow for ad-hoc "injection" of user input - definitely interesting
Probably will address #219
Instead of having human input only via agent making a tool call and the user input in the response, instruct the agents that user input may come at any time in the thread, but they must use a tool to specifically request a response from the user.
The pattern the agent would see is:
The key is we have to show the tool call so that agents don't conclude from history that they can converse with users by posting "regular" messages.
This will also facilitate adding user messages at any point, perhaps with a mechanism like
sending a global message doesn't seem necessary at the moment.
The text was updated successfully, but these errors were encountered: