Install the gateway

The Inviolet gateway is the choke point every LLM tool call passes through. Install one of three ways depending on where your application lives.

Option 1 — npm SDK (simplest)

Guard tool calls from your existing app with the Inviolet client. No infrastructure changes.

npm install @inviolet/agent-sdk-core

import { Inviolet } from "@inviolet/agent-sdk-core"

const inviolet = new Inviolet({
  gatewayUrl: process.env.INVIOLET_GATEWAY_URL!,
  apiKey: process.env.INVIOLET_API_KEY!,
  agentId: "my-app",
})
// Mint a mandate, then guard each tool call — see the Quickstart.

If your agent uses a framework, install its wrapper instead (Anthropic, OpenAI, LangChain, and more) — the same Inviolet client underneath. Best for: serverless apps, Next.js, single-team prototypes.

Option 2 — Docker (self-hosted)

docker run -d \
  -e INVIOLET_API_KEY=${INVIOLET_API_KEY} \
  -p 8080:8080 \
  ghcr.io/craig-fire/inviolet-gateway:latest

Point your application at http://localhost:8080 instead of the LLM provider’s API URL. Best for: regulated environments, on-prem, self-hosted Kubernetes.

Option 3 — Vercel Edge

Deploy the gateway as an edge function alongside your Next.js app.

// app/api/llm/route.ts
import { invioletEdge } from "@inviolet/edge"
export const runtime = "edge"
export const POST = invioletEdge({ apiKey: process.env.INVIOLET_API_KEY! })

Best for: existing Vercel-hosted apps that want zero new infrastructure.

Verify install

After install, make any LLM tool call. Within 30 seconds it should appear at app.inviolet.ai/decision-feed.

Install & configure

Deploy via MDM

Define + enforce

Operate

Install the gateway

Option 1 — npm SDK (simplest)

Option 2 — Docker (self-hosted)

Option 3 — Vercel Edge

Verify install

Read next

​Option 1 — npm SDK (simplest)

​Option 2 — Docker (self-hosted)

​Option 3 — Vercel Edge

​Verify install

​Read next

Option 1 — npm SDK (simplest)

Option 2 — Docker (self-hosted)

Option 3 — Vercel Edge

Verify install

Read next