[Bug]: LangGraph agents in Foundry Hosted agents are not emitting gen_ai.agent.* metadata

[shivamMg]

Component

Langchain

Description

LangGraph/LangChain agents that are being run inside Microsoft Foundry as Hosted agents are not emitting Agent name, version and id in the right fields.

Instead of emitting them in the fields dictated by OTel GenAI Semconv. Instead of these fields:

gen_ai.agent.name
gen_ai.agent.version
gen_ai.agent.id

The following fields are being set:

microsoft.gen_ai.main_agent.name
microsoft.gen_ai.main_agent.version
microsoft.gen_ai.main_agent.id

Expected Behavior

The following fields are not being set for langgraph agents:

gen_ai.agent.name
gen_ai.agent.version
gen_ai.agent.id

Steps to Reproduce

Try the sample LangGraph agent from: https://github.com/shivamMg/fine-tuning/tree/d49b6e462db7f3795a7eb8984aa8900a03780b42/Demos/AgentTracesToSFT/retail-agent-langgraph

Environment

Microsoft Foundry Hosted agents

Additional Context

No response

[github-actions]

cc @hectorhdzg @JacksonWeber @lzchen @MSNev @rads-1996 @jeremydvoss @nagkumar91

[rads-1996]

@shivamMg I think I know what is wrong here, in your sample when you invoke the microsoft distro, you pass the agent_id, name and description as config parameters to the distro instead of passing it in the instrumentation options for langchain -

You should instead be passing these values as follows -

use_microsoft_opentelemetry(
    enable_azure_monitor=True,
    sampling_ratio=1.0,
    instrumentation_options={
        "langchain": {
            "enabled": True,
            # Optional: set static agent identity on all LangChain spans.
            # These appear as gen_ai.agent.* attributes in your telemetry.
            # If omitted, agent_name is inferred from LangChain's
            # runtime metadata when available.
            agent_name=os.getenv("AGENT_NAME", "retail-agent-langgraph"),
            agent_version=os.getenv("AGENT_VERSION"),
            agent_id=os.getenv("AGENT_ID"),
        },
    },
    enable_sensitive_data=True,
)

Please try out the correct format and let us know if the issue persists.

[github-actions]

@shivamMg we need more information before we can continue working on this issue.

Please reply within 14 days. If we do not hear back, this issue may be closed automatically.

Marked as needs-author-feedback on 2026-07-16.

[github-actions]

@shivamMg friendly reminder: we are still waiting for your feedback.

If we do not hear back within 7 more days, this issue may be closed automatically.