Add run configuration inheritance for agent-as-tool workflows #1006
+28
−14
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
I am not sure what @rh-openai thinks but here are my initial thoughts on this:
|
rm-openai
requested changes
Jul 8, 2025
There was a problem hiding this comment.
Yeah using the trace to store this data isn't right. What you could do instead, is use a ContextVar. So:
- Add a
pass_run_config_to_sub_agents: boolparam to the RunConfig, just in case people want to enable/disable this behavior - Set the run_config context var from run.py
- Read the run_config in the agent as tool
- Also ensure that you add tests to make sure the contextvar is unset in all cases (errors, success, cancel etc)
I'm also not entirely sure if it makes sense to have this as part of the SDK, or just something you do. The approach I've taken to as_tool is that if you want customization, you can just create your own tool via:
@function_tool
def run_my_agent(input):
my_agent = Agent(...)
result = Runner.run(my_agent)
return result.foo
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Problem:
When using the "agents as tools" pattern, tool-agents create their own execution context but don't inherit the parent's run_config.
This creates two problems:
Tool-agents can't access the parent's global model configuration
If tool-agents don't have their own model field set and the parent's run_config.model is unavailable, model resolution fails entirely, causing the orchestrator to fail when calling the tool
Solution:
Modified the agent-as-tool execution to properly handle run_config inheritance while ensuring robust model resolution. This prevents execution failures when tool-agents lack explicit model configuration.
Changes
1. traces.py
_run_configattribute to bothNoOpTraceandTraceImplclassesTYPE_CHECKINGimport forRunConfigtype annotation2. run.py
_run_configattribute (fallback)3. agent.py
get_current_traceimport from tracing moduleas_tool()method to retrieve and inherit run configuration from trace contextRunner.run()for consistent behavior