|
| 1 | +from __future__ import annotations |
| 2 | + |
| 3 | +import asyncio |
| 4 | + |
| 5 | +from openai.types.responses import ResponseTextDeltaEvent |
| 6 | +from pydantic import BaseModel, Field |
| 7 | + |
| 8 | +from agents import Agent, Runner |
| 9 | + |
| 10 | +""" |
| 11 | +This example shows how to use guardrails as the model is streaming. Output guardrails run after the |
| 12 | +final output has been generated; this example runs guardails every N tokens, allowing for early |
| 13 | +termination if bad output is detected. |
| 14 | +
|
| 15 | +The expected output is that you'll see a bunch of tokens stream in, then the guardrail will trigger |
| 16 | +and stop the streaming. |
| 17 | +""" |
| 18 | + |
| 19 | + |
| 20 | +agent = Agent( |
| 21 | + name="Assistant", |
| 22 | + instructions=( |
| 23 | + "You are a helpful assistant. You ALWAYS write long responses, making sure to be verbose " |
| 24 | + "and detailed." |
| 25 | + ), |
| 26 | +) |
| 27 | + |
| 28 | + |
| 29 | +class GuardrailOutput(BaseModel): |
| 30 | + reasoning: str = Field( |
| 31 | + description="Reasoning about whether the response could be understood by a ten year old." |
| 32 | + ) |
| 33 | + is_readable_by_ten_year_old: bool = Field( |
| 34 | + description="Whether the response is understandable by a ten year old." |
| 35 | + ) |
| 36 | + |
| 37 | + |
| 38 | +guardrail_agent = Agent( |
| 39 | + name="Checker", |
| 40 | + instructions=( |
| 41 | + "You will be given a question and a response. Your goal is to judge whether the response " |
| 42 | + "is simple enough to be understood by a ten year old." |
| 43 | + ), |
| 44 | + output_type=GuardrailOutput, |
| 45 | + model="gpt-4o-mini", |
| 46 | +) |
| 47 | + |
| 48 | + |
| 49 | +async def check_guardrail(text: str) -> GuardrailOutput: |
| 50 | + result = await Runner.run(guardrail_agent, text) |
| 51 | + return result.final_output_as(GuardrailOutput) |
| 52 | + |
| 53 | + |
| 54 | +async def main(): |
| 55 | + question = "What is a black hole, and how does it behave?" |
| 56 | + result = Runner.run_streamed(agent, question) |
| 57 | + current_text = "" |
| 58 | + |
| 59 | + # We will check the guardrail every N characters |
| 60 | + next_guardrail_check_len = 300 |
| 61 | + guardrail_task = None |
| 62 | + |
| 63 | + async for event in result.stream_events(): |
| 64 | + if event.type == "raw_response_event" and isinstance(event.data, ResponseTextDeltaEvent): |
| 65 | + print(event.data.delta, end="", flush=True) |
| 66 | + current_text += event.data.delta |
| 67 | + |
| 68 | + # Check if it's time to run the guardrail check |
| 69 | + # Note that we don't run the guardrail check if there's already a task running. An |
| 70 | + # alternate implementation is to have N guardrails running, or cancel the previous |
| 71 | + # one. |
| 72 | + if len(current_text) >= next_guardrail_check_len and not guardrail_task: |
| 73 | + print("Running guardrail check") |
| 74 | + guardrail_task = asyncio.create_task(check_guardrail(current_text)) |
| 75 | + next_guardrail_check_len += 300 |
| 76 | + |
| 77 | + # Every iteration of the loop, check if the guardrail has been triggered |
| 78 | + if guardrail_task and guardrail_task.done(): |
| 79 | + guardrail_result = guardrail_task.result() |
| 80 | + if not guardrail_result.is_readable_by_ten_year_old: |
| 81 | + print("\n\n================\n\n") |
| 82 | + print(f"Guardrail triggered. Reasoning:\n{guardrail_result.reasoning}") |
| 83 | + break |
| 84 | + |
| 85 | + |
| 86 | +if __name__ == "__main__": |
| 87 | + asyncio.run(main()) |
0 commit comments