Documentation — OmniaGuard | AI Cybersecurity Platform

Quickstart

The fastest path to protection is the inline API gateway. Route your existing AI API calls through OmniaGuard's endpoint — your code changes by exactly one URL.

Step 1 — Get your API Key

After signing up, your API key is available on your dashboard. All keys follow the format og_live_xxxxxxxxxxxx.

Step 2 — Route through the gateway

Before (direct OpenAI)
const response = await openai.chat.completions.create({
  model: "gpt-4",
  messages: [{ role: "user", content: userInput }]
});

After (OmniaGuard protected)
const response = await fetch("https://api.omniaguard.ai/v1/proxy", {
  method: "POST",
  headers: {
    "Content-Type": "application/json",
    "X-OmniaGuard-Key": "og_live_xxxxxxxxxxxx",
    "X-Target-Model": "openai/gpt-4"
  },
  body: JSON.stringify({
    messages: [{ role: "user", content: userInput }]
  })
});

That's it. All 14 agents are now inspecting every request before it reaches OpenAI. Zero changes to your downstream code.

Authentication

OmniaGuard uses API key authentication. Include your key in the X-OmniaGuard-Key header on every request.

Environment Variable (recommended)
# .env — never commit this file
OMNIAGUARD_API_KEY=og_live_xxxxxxxxxxxx

Keys are scoped to your account. Enterprise customers can create per-project sub-keys with granular permissions from the dashboard.

SDK Installation

Node.js / TypeScript

npm

npm install @omniaguard/sdk

TypeScript
import { OmniaGuard } from '@omniaguard/sdk';

const og = new OmniaGuard({
  apiKey: process.env.OMNIAGUARD_API_KEY,
  model: 'openai/gpt-4',
  agents: 'all' // or ['prime', 'vault', 'forge']
});

const result = await og.inspect({
  messages: [{ role: 'user', content: userInput }]
});

if (result.safe) {
  // proceed to your AI model
} else {
  console.log('Threat blocked:', result.threat);
}

Python

pip
pip install omniaguard

Python
from omniaguard import OmniaGuard

og = OmniaGuard(api_key=os.environ["OMNIAGUARD_API_KEY"])

result = og.inspect(
    messages=[{"role": "user", "content": user_input}],
    agents="all"
)

if not result.safe:
    raise SecurityError(result.threat_type)

API Reference

The OmniaGuard API is RESTful. All requests use HTTPS. Base URL: https://api.omniaguard.ai/v1

Endpoints

POST/proxyInline proxy — inspect and forward to any AI model endpoint

POST/inspectInspect only — return threat verdict without forwarding

GET/agents/statusGet real-time status of all 14 agents

GET/threatsList recent blocked threats with forensic details

GET/auditFetch audit log entries (paginated, filterable)

POST/policyCreate or update a custom security policy

GET/usageRequest counts, blocked threats, and quota usage

Agent Configuration

By default all agents on your tier are active. You can customize which agents inspect each request type using policy objects.

Agent Policy JSON
{
  "policy_name": "production",
  "agents": {
    "prime": { "enabled": true, "sensitivity": "high" },
    "vault": { "enabled": true, "pii_redaction": true },
    "forge": { "enabled": true, "rate_limit": 1000 },
    "nexus": { "quorum": 7 }
  },
  "block_action": "reject",
  "log_blocked": true
}

Error Codes

OG-001 — Prompt injection detected (blocked by Agent Prime)
OG-002 — Jailbreak attempt detected (blocked by Agent Bastion)
OG-003 — Credential exposure risk (blocked by Agent Vault)
OG-004 — Rate limit exceeded (blocked by Agent Forge)
OG-005 — Consensus threshold not met (blocked by Agent Nexus)
OG-010 — Authentication failed (invalid API key)
OG-020 — Quota exceeded (upgrade plan for more requests)

OmniaGuard Docs