Vulnetix AI Firewall

Documentation

User documentation for the Vulnetix AI Firewall — the gateway, its guardrails, and every client you can put in front of it.

The AI Firewall is a gateway at guardrails.vulnetix.com that sits between your LLM clients and the providers they call. Every request is authenticated to your organisation, checked against your provider and model policy and your content guardrails, and only then forwarded — with your own provider key, decrypted just-in-time from a vault the gateway can read but you cannot.

Getting Started
Base URLs, the two keys people confuse, and the BYOK vault.
SDK Tutorials
Python, TypeScript, Go, Java, C#, Ruby, PHP, Rust, and more.
Coding Agents
Claude Code, Claude Desktop, Codex CLI, Cursor, Continue, Aider, OpenCode.
Guardrails
Block, redact, flag. Secrets, PII, prompt injection, conversation caps.
Providers & Models
Allow and deny vendors and models across the whole organisation.
Sovereign AI
Zero data retention, key custody, and the threat model.
Using the CLI
Wire your machine, check your posture, manage policy as code.
Reference
Endpoints, the full error taxonomy, troubleshooting, limitations.

What it is not

Being clear about this up front saves a support ticket later. The AI Firewall has no rate limits, no spend caps, and no budgets. It does not mark up your tokens or resell them — you keep your own billing relationship with each provider. It does not inspect completions, only the requests you send. And it is a control, not a cage: a developer who unsets an environment variable is talking to the provider directly again, which is why vulnetix ai-firewall status exists.

See Limitations & roadmap for the full list.

Coding Agents
Put Claude Code, Claude Desktop, Codex CLI, Cursor, Continue, Aider and every other OpenAI-compatible agent behind the firewall — and know exactly which of their traffic does not go through it.
Getting Started
Get your first firewalled LLM call working, and understand the two keys, the base URLs, and the BYOK vault before you wire anything else.
SDK Tutorials
Point any LLM client at the AI Firewall. Two settings change — base_url and api_key — and nothing else about your code does.
Guardrails
Content rules applied to every request the firewall proxies — block secrets, redact PII, cap runaway agent loops. Three rule types, three actions, evaluated in priority order.
Providers & Models
Decide which AI vendors your organisation may use, and which models — enforced on every request, for every client.
Sovereign AI
Keep control of your data, your keys, and which jurisdictions your prompts reach — with an honest account of what the firewall guarantees and what it cannot.
Using the CLI
Wire your machine, check your posture, and manage policy as code. Task-oriented — the full flag reference lives in the CLI's own docs.
Reference
Endpoints, the full error taxonomy, troubleshooting, alternatives, and an honest account of what the product does not do.