Eduard van Valkenburg
ce74c84bdb
* Python: Preserve OTel parent context for deferred streams - capture current OTel context when opening host-managed streaming runs - re-activate captured context during deferred stream pulls and finalization - add host-level regression coverage for deferred stream parent-span linkage - add Responses channel integration coverage for request parent span propagation Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Python: Capture OTel stream context before target.run - capture OTel context snapshot before invoking target.run in _invoke_stream - add regression test guarding capture-before-run evaluation order Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
agent-framework-hosting-responses
OpenAI Responses-shaped channel for agent-framework-hosting.
Exposes a single POST /responses endpoint that accepts the OpenAI
Responses API request body and returns either a Responses-shaped JSON
body or a Server-Sent-Events stream when stream=True.
from agent_framework.openai import OpenAIChatClient
from agent_framework_hosting import AgentFrameworkHost
from agent_framework_hosting_responses import ResponsesChannel
agent = OpenAIChatClient().as_agent(name="Assistant")
host = AgentFrameworkHost(target=agent, channels=[ResponsesChannel()])
host.serve(port=8000)
The base host plumbing lives in
agent-framework-hosting.