Repository navigation
Passing output of one agent to another in a background job #458
Replies: 1 comment
|
Your use of a distinct Two ADK details help clarify the design:
A practical structure for your example is:
In other words, preserve your existing stable job IDs and map them to session output keys; The source links above are from current ADK Go main (v2 development). This is an architecture recommendation based on source inspection, not a reproduction of your snippet. Which ADK version and session storage implementation are you using? |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Hey I'm currently implementing a use case where an Agent uses a StructuredOutput, returns a 200 OK back to the client and starts a background job using the output parameters. This background job has a new parallel agent that utilizes the StructuredOutput the previous agent as InputSchema. A simplified snippet:
I'm looking to preserve the ids of the jobs, so the output can be accessed later. Is there a more idiomatic way to do this nonblocking call (e.g. through some agent pattern) that still preserves the unique ID of jobs so I am able to distinguish the outputs?
All reactions