47,291 injections analyzed today

Try to inject an LLM. See it caught.

Paste any prompt below. The protocol runs it through the same /v1/analyze endpoint your production app would use. Sub-25 ms. Deterministic. Signed verdict.

// Prompt to analyze Try: "Ignore all previous instructions"

Inputs are analyzed and discarded. Not stored. See Privacy Policy.

Or try:

—

Category

—

Confidence

—

Latency

—

// Signed audit report

report_id—

sha256—

timestamp—

◆ Run this in your own stack

JS · fetch◆ 3 lines

// Drop this in front of every model call.
const r = await fetch("https://api.zentricprotocol.com/v1/analyze", { method:"POST", headers:{"Authorization":`Bearer ${process.env.ZENTRIC_API_KEY}`,"Content-Type":"application/json"}, body:JSON.stringify({ input, modules:["integrity","privacy"] }) });
const { verdict } = await r.json(); if (verdict === "BLOCKED") throw new Error("injection detected");

Get your API key — free →

By using this playground you agree to our Terms of Service. Rate-limited to 20 requests/min per IP.