mirror of
https://github.com/jillesca/oncall-netops.git
synced 2026-11-19 02:13:48 +00:00
62 lines
2.0 KiB
Python
62 lines
2.0 KiB
Python
from langchain_openai import ChatOpenAI
|
|
from langgraph.prebuilt import create_react_agent
|
|
from tools.pyats.pyats_interface_oper_tools import (
|
|
shut_interface,
|
|
unshut_interface,
|
|
)
|
|
from tools.pyats.pyats_interface_status_tools import (
|
|
verify_interface_state_up,
|
|
)
|
|
from tools.pyats.pyats_interface_status_tools import (
|
|
get_interfaces_status,
|
|
)
|
|
|
|
|
|
AGENT_INTERFACE_ACTIONS = """You are a network tech lead that is specialized in operating interfaces of network devices. You have access to a set of tools to connect to network devices and perform actions on the interfaces. You decide tool or tools to use.
|
|
|
|
Validate the interface name given to you is real, use the gget_interfaces_status tool to check if the interface exists. This is a mandatory step.
|
|
|
|
You will be given a list of interfaces and the action to be performed on them. You must use the tools to perform the action on the interfaces.
|
|
|
|
Once you have the correct device name, perform the action asked on the interfaces and always verify your work with the verify_interface_state_up tool.
|
|
"""
|
|
|
|
LLM_MODEL = "gpt-4o-mini"
|
|
tools = [
|
|
shut_interface,
|
|
unshut_interface,
|
|
verify_interface_state_up,
|
|
get_interfaces_status,
|
|
]
|
|
model = ChatOpenAI(model=LLM_MODEL, temperature=0)
|
|
|
|
|
|
def agent_interface_actions():
|
|
return create_react_agent(
|
|
model, tools=tools, prompt=AGENT_INTERFACE_ACTIONS
|
|
)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
inputs = {
|
|
"messages": [
|
|
# (
|
|
# "user",
|
|
# "Can you shut the interface GigabitEthernet4 on device cat8000v-0?",
|
|
# ),
|
|
(
|
|
"user",
|
|
"Can you unshut the interface GigabitEthernet4 on device cat8000v-0?",
|
|
),
|
|
]
|
|
}
|
|
|
|
graph = agent_interface_actions()
|
|
|
|
for s in graph.stream(inputs, stream_mode="values"):
|
|
message = s["messages"][-1]
|
|
if isinstance(message, tuple):
|
|
print(message)
|
|
else:
|
|
message.pretty_print()
|