ProductsGateway

Guardrails

Catch prompt injections, jailbreaks, system prompt leaks, and unsafe content before they reach your models and tools. Built on Span-01, free with the Respan gateway.

Respan Guardrails are in early access. Join the waitlist to turn them on for your organization.

Respan Guardrails check every request that goes through the Respan gateway with Span-01, Respan’s classification model. Turn them on once, pick the categories you care about, and threats are caught before they reach your model.

What Guardrails check

Choose any combination of these categories, or select all of them.

GroupCategory
SecurityJailbreak / prompt injection
System prompt leak
Content safetyUnsafe content
Hate / harassment
Self-harm
Sexual content
Violence
Illegal instructions

Benchmarks

On a benchmark built from 7 public security datasets, Respan Guardrails post the highest average F1 of the guardrails we tested, and they’re completely free with the Respan gateway.

Guardrail quality versus cost. Average F1 across 8 guardrail categories: Span-01 0.637 at $0, TypeSafe Jev 0.627, gpt-oss safeguard 20B 0.621 at an estimated $0.85 per 1,000 checks, Shieldstral 3B 0.590, OpenAI Moderation 0.481 at $0, Azure AI Content Safety 0.377 at $0.89 per 1,000 checks.
Average F1 across the 8 guardrail categories. * Estimated hosted cost.

Setup

1

Create a Respan account

Sign up on Respan, then create an API key on the API keys page.

2

Route your app through the Respan gateway

Guardrails run inside the Respan gateway, so they check each request inline, before it reaches your model. The Respan CLI hands off to your coding agent (Claude Code, Cursor, Codex, and others), which points your project at the gateway for you.

npx @respan/cli setup gateway

To set up the gateway by hand, see the gateway quickstart.

Guardrails apply only to requests sent through the Respan gateway. Apps that send traces to Respan without the gateway are not covered.

3

Choose the categories to block

Open the Guardrails page in the Respan platform. Under Blocked categories, open Select categories and choose the categories to block, or pick Select all. No code changes needed.

Gateway requests that match a selected category are blocked, and show up under Guardrail activity.

Guardrails page in the Respan platform. Under Span-01 Guardrails, the Blocked categories menu lists Jailbreak / prompt injection, System prompt leak, Unsafe content, Hate / harassment, Self-harm, Sexual content, Violence, and Illegal instructions. Guardrail activity is shown below.