OpenAI

pip install humanlayer openai python-dotenv

from humanlayer import HumanLayer
from openai import OpenAI
import json
from dotenv import load_dotenv

load_dotenv()

hl = HumanLayer(
    verbose=True,
    run_id="openai-math-example"
)

def add(x: int, y: int) -> int:
    """Add two numbers together."""
    return x + y

@hl.require_approval()
def multiply(x: int, y: int) -> int:
    """multiply two numbers"""
    return x * y

# Define tools for OpenAI
math_tools = [
    {
        "type": "function",
        "function": {
            "name": "multiply",
            "description": "multiply two numbers",
            "parameters": {
                "type": "object",
                "properties": {
                    "x": {"type": "number"},
                    "y": {"type": "number"},
                },
                "required": ["x", "y"],
            },
        },
    },
    {
        "type": "function",
        "function": {
            "name": "add",
            "description": "add two numbers",
            "parameters": {
                "type": "object",
                "properties": {
                    "x": {"type": "number"},
                    "y": {"type": "number"},
                },
                "required": ["x", "y"],
            },
        },
    },
]

def run_chain(prompt: str, tools: list[dict], tools_map: dict) -> str:
    client = OpenAI()
    messages = [{"role": "user", "content": prompt}]

    response = client.chat.completions.create(
        model="gpt-4",
        messages=messages,
        tools=tools,
        tool_choice="auto",
    )

    while response.choices[0].finish_reason != "stop":
        response_message = response.choices[0].message
        tool_calls = response_message.tool_calls
        if tool_calls:
            messages.append(response_message)
            for tool_call in tool_calls:
                function_name = tool_call.function.name
                function_to_call = tools_map[function_name]
                function_args = json.loads(tool_call.function.arguments)

                try:
                    function_response = function_to_call(**function_args)
                    function_response_json = json.dumps(function_response)
                except Exception as e:
                    function_response_json = json.dumps({"error": str(e)})

                messages.append({
                    "tool_call_id": tool_call.id,
                    "role": "tool",
                    "name": function_name,
                    "content": function_response_json,
                })

        response = client.chat.completions.create(
            model="gpt-4",
            messages=messages,
            tools=tools,
        )

    return response.choices[0].message.content

from humanlayer.core.models import FunctionCallSpec

approval = hl.fetch_approval(
    FunctionCallSpec(
        fn="multiply",
        kwargs={"x": 2, "y": 5}
    ),
)

completed = approval.as_completed()
if completed.approved:
    result = multiply(x=2, y=5)
else:
    print(f"Rejected with comment: {completed.comment}")

call = hl.create_function_call(
    spec=FunctionCallSpec(
        fn="multiply",
        kwargs=function_args,
    ),
    call_id=tool_call.id,  # Optional: use OpenAI's tool_call_id
)

while (not call.status) or (call.status.approved is None):
    time.sleep(5)
    call = hl.get_function_call(call_id=tool_call.id)

if call.status.approved:
    result = multiply(**function_args)
else:
    print(f"Rejected: {call.status.comment}")