Version
1.7.x
Which installation method(s) does this occur on?
PyPi
Describe the bug.
front_end_plugin_worker.py:218 unconditionally accesses workflow.config.workflow.workflow_alias, but LanggraphWrapperConfig extends FunctionBaseConfig, not AgentBaseConfig — so the attribute doesn't exist and you get an AttributeError.
The goal was to wrap my LangGraph agent with NAT and then expose the agent as an MCP tool call. The goal was NOT to expose the tools inside of my LangGraph graph, just the agent.
Minimum reproducible example
workflow.yaml
llms:
agent:
_type: nim
model: meta/llama-3.1-8b-instruct
api_key: dummy
functions:
current_datetime:
_type: current_datetime
workflow:
_type: langgraph_wrapper
llm_name: agent
graph: some.module:graph
general:
front_end:
_type: mcp
host: 0.0.0.0
port: 9901
Relevant log output
Click here to see error details
AttributeError: 'LanggraphWrapperConfig' object has no attribute 'workflow_alias'
Other/Misc.
No response
Code of Conduct
Version
1.7.x
Which installation method(s) does this occur on?
PyPi
Describe the bug.
front_end_plugin_worker.py:218 unconditionally accesses workflow.config.workflow.workflow_alias, but LanggraphWrapperConfig extends FunctionBaseConfig, not AgentBaseConfig — so the attribute doesn't exist and you get an AttributeError.
The goal was to wrap my LangGraph agent with NAT and then expose the agent as an MCP tool call. The goal was NOT to expose the tools inside of my LangGraph graph, just the agent.
Minimum reproducible example
workflow.yaml llms: agent: _type: nim model: meta/llama-3.1-8b-instruct api_key: dummy functions: current_datetime: _type: current_datetime workflow: _type: langgraph_wrapper llm_name: agent graph: some.module:graph general: front_end: _type: mcp host: 0.0.0.0 port: 9901Relevant log output
Click here to see error details
Other/Misc.
No response
Code of Conduct