Production guardrails, one slug away.

Firewall sits before the model. Group rules, pick a mode (signal_only or blocked), and reference the slug in any OpenAI-compatible request.

Start with Compute See built-in rules

OpenAI-compatible

response = client.chat.completions.create(
    model="your-model",
    messages=[{"role": "user", "content": prompt}],
    extra_body={"firewall_slug": "production-guardrails"},
)

See the request path before the model sees anything.

The firewall evaluates every request before inference starts. Unsafe input stops at the boundary. Safe input keeps moving.

User

Shopper

Message

f*** you

Firewall

Policy check

blocked

403 returned before inference

Assistant

Model

Response

No response sent

User

Shopper

Message

Hey!

Firewall

Policy check

allowed

Request reaches the model

Assistant

Model

Response

I'm doing well — how can I help?

Two ways to enforce policy.

Warn the model. Let it explain.

Model sees the warning

The firewall evaluates the request. If it flags something, it tells the model the request is unsafe and instructs it to explain why carefully to the user. The model still responds, but with guardrail context.

Stop unsafe requests before inference.

403 — model never called

The firewall evaluates the request. If it flags something, it returns a 403 before the model ever sees it. No compute wasted, no unsafe response possible.

Production rules, maintained for you.

QDivZero ships with seven guardrail rules across safety, privacy, abuse, and security categories. They are updated as threats evolve.

Rule intelligence

1/7

abuse

high

Malware, fraud, and abuse

public:malware-fraud-abuse

Flags requests about malware, scams, fraud, or abusive misuse.

Your rules, your prompts.

Build rules with your own prompts for what your application needs to filter. Same modes, same API, same enforcement path as built-in rules.

“Don't let the model talk about football.”

Custom rules use the same signal_only or blocked mode as built-in rules. Describe what to catch and the firewall handles the rest.

What Firewall already includes.

Built-in rules

Seven production-maintained guardrail rules.

Custom rules

Define your own policies with prompts.

signal_only mode

Warn the model so it can explain.

blocked mode

Return 403 before inference starts.

OpenAI-compatible

Pass firewall as a request parameter.

Evaluator model

Configurable model that runs the evaluation.

Attach a firewall slug to any request.

Pass a firewall slug to any OpenAI-compatible request. The evaluator runs before the model, and the mode decides what happens next.

View docs

quickstart.py

from openai import OpenAI

client = OpenAI(
    base_url="https://api.qdiv0.com/v1",
    api_key="your-api-key",
)

response = client.chat.completions.create(
    model="your-model",
    messages=[
        {"role": "user", "content": prompt},
    ],
    extra_body={"firewall_slug": "production-guardrails"},
)

Add guardrails to production inference.

Start with Compute, configure a firewall, and keep unwanted questions away from your models.

Start with Compute Access the platform