Example use cases

import os
from openai import OpenAI
 
client = OpenAI(
    api_key=os.environ.get("OPENAI_API_KEY"),
    base_url="https://api.grayswan.ai/cygnal",
    default_headers={
        "grayswan-api-key": os.environ.get("GRAYSWAN_API_KEY"),
        # Set legal assistance boundaries
        "category-legal-advice": "Avoid providing specific legal advice that could constitute unauthorized practice of law",
        "category-disclaimer": "Always clarify that information is for educational purposes only and not a substitute for professional legal advice"
    }
)
 
# Legal information query
completion = client.chat.completions.create(
    model="gpt-4.1",
    messages=[
        {"role": "user", "content": "How do I file for bankruptcy in my situation? I owe $50,000 in credit card debt."}
    ],
)
 
print(completion.choices[0].message.content)
# Will provide general information with appropriate disclaimers

import os
from openai import OpenAI
 
client = OpenAI(
    api_key=os.environ.get("OPENAI_API_KEY"),
    base_url="https://api.grayswan.ai/cygnal",
    default_headers={
        "grayswan-api-key": os.environ.get("GRAYSWAN_API_KEY"),
        # Set cybersecurity boundaries
        "category-malicious-code": "Prevents generation of exploits, malware, or other harmful code",
        "category-responsible-disclosure": "Encourages ethical security practices and responsible disclosure",
        "category-attack-prevention": "Blocks instructions for launching cyber attacks"
    }
)
 
# Security researcher query
completion = client.chat.completions.create(
    model="gpt-4.1",
    messages=[
        {"role": "user", "content": "I found a SQL injection vulnerability. How can I demonstrate this to the company securely?"}
    ],
)
 
print(completion.choices[0].message.content)
# Will provide responsible disclosure guidance without generating actual exploits

import os
from openai import OpenAI
 
# Initialize the client with Cygnal
client = OpenAI(
    api_key=os.environ.get("OPENAI_API_KEY"),
    base_url="https://api.grayswan.ai/cygnal",
    default_headers={
        "grayswan-api-key": os.environ.get("GRAYSWAN_API_KEY"),
        # Set education-specific categories
        "category-direct-answers": "No direct solutions to homework problems without guiding the student through the thought process",
        "category-educational-guidance": "Provide step-by-step guidance, explanations of concepts, and Socratic questioning to help students learn"
    }
)
 
# Example student query about a homework problem
completion = client.chat.completions.create(
    model="gpt-4.1",
    messages=[
        {"role": "user", "content": "I need to solve this equation for homework: 2x + 5 = 13. What's the answer?"}
    ],
)
 
print(completion.choices[0].message.content)
# The model will guide the student through solving the equation instead of just giving "x = 4"