BrokenGPT Python API quickstart

1. Create an API key in the BrokenGPT dashboard and copy it once.
2. Install a current OpenAI-compatible Python client with pip install openai.
3. Set BROKENGPT_API_KEY in the server process or secret manager, never in committed code.

# PowerShell
$env:BROKENGPT_API_KEY="your-key"

# bash / zsh
export BROKENGPT_API_KEY="your-key"

The base URL ends at /v1 and the public model alias is broken-one. A non-streamed call is the easiest way to validate authentication, request fields, and response parsing.

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["BROKENGPT_API_KEY"],
    base_url="https://brokengpt.com/v1",
)

response = client.chat.completions.create(
    model="broken-one",
    messages=[{"role": "user", "content": "Explain SSE in three sentences."}],
)

print(response.choices[0].message.content)

The chat-completions route streams server-sent events. Your client should tolerate empty deltas, inspect finish reasons, stop cleanly on completion, and handle a connection that closes before the terminal event.

stream = client.chat.completions.create(
    model="broken-one",
    messages=[{"role": "user", "content": "Write a short launch checklist."}],
    stream=True,
)

for chunk in stream:
    text = chunk.choices[0].delta.content if chunk.choices else None
    if text:
        print(text, end="", flush=True)

• Do not retry malformed requests or invalid credentials.
• Retry 429 and temporary 5xx responses with exponential backoff, jitter, and a maximum attempt count.
• Log the returned request ID without logging the API key or sensitive prompt content.
• Validate tool arguments and structured output before using them.
• Set application timeouts and propagate cancellation when the caller disconnects.