{graph_notation}

{graph_spec}

Here is the python State class associated with this graph.

{state_code}

Nodes are specified in markdown files in sections for each node.

BEHAVIOR: the comment above the node specifies the behavior
USER_PROMPT: prompt for user, what we are asking for, if anything
GENERATION_PROMPT: prompt for model to generate data
READS: field_name_1
WRITES: field_name_3, field_name_7

Here is the node specification:

{node_spec}

Please do the following tasks:

1. Write the Routing Functions for this graph
2. Write the langgraph builder code for this graph, see examples below
3. Write this code to the file 'graph_code.py'

• at the top of the file, add comment indicating file was generated using {model_name}
• import the graph state from state_code.py
• import the node functions from node_code.py

4. Do not write any Node Functions, only write Routing Functions and the Graph Builder code

Here's an example graph:

from langchain_core.runnables.config import RunnableConfig from langgraph.graph import START, END, StateGraph from langgraph.checkpoint.sqlite import SqliteSaver

def make_builder(): builder = StateGraph(JokesterState)

# Add node functions
builder.add_node('get_joke_topic', get_joke_topic)
builder.add_node('tell_joke', tell_joke)
builder.add_node('ask_for_another', ask_for_another)

# Add edges for execution flow
builder.add_edge(START, 'get_joke_topic')
builder.add_edge('get_joke_topic', 'tell_joke')
builder.add_edge('tell_joke', 'ask_for_another')

# Conditional routing from ask_for_another node
# Note the special handling for END, we automatically map the string 'END' to the constant END.
# This means the Routing Function returns the string values 'tell_joke' and 'END', and these map to 'tell_joke' and END (the constant, special case only for 'END')
conditional_edges = ⟨ 'tell_joke': 'tell_joke', 'END': END ⟩
builder.add_conditional_edges('ask_for_another', tell_another, conditional_edges)
return builder

NOTE: Worker Nodes require special handling in the graph.

from langchain_core.runnables.config import RunnableConfig from langgraph.graph import START, END, StateGraph

In this example, we have:

#. GRAPH_LINE: orchestrator -> llm_call(State.sections)

The "State.sections" is the iterable, and the

from langgraph.constants import Send

def assign_workers(state: State): """Assign a worker to each section in the plan"""

# Kick off section writing in parallel via Send() API
return [Send("llm_call", ⟨ "section": s⟩) for s in state["sections"]]

building graph like previous example, this shows how to transition to Worker Node

to add that transition to the graph

orchestrator_worker_builder.add_conditional_edges( "orchestrator", assign_workers, ["llm_call"] )

Please do the following tasks:

1. Write the Routing Functions for this graph
2. Write the langgraph builder code for this graph. Only write the function that returns the builder. Do not compile the graph or call this function. That code will be written elsewhere. Only write this function. Name it 'make_builder'
3. Write this code to the file 'graph_code.py'

• at the top of the file, add comment indicating file was generated using {model_name}
• import the graph state from state_code.py
• import the node functions from node_code.py

4. Do not write any Node Functions, only write Routing Functions and the Graph Builder code

Write the Routing Functions and Builder code to file: graph_code.py.

{question}