Developers

AI Application Firewall API

The AI Application Firewall API brings the same checks to applications that don’t run on Autessa: POST each phase of a turn to /clients/aaf/inspect and get back ALLOW, FLAG, CHALLENGE, REQUIRE_APPROVAL, BLOCK or DURESS with reasons and findings.

Turn on the firewall

RequestSAMPLE

POST https://api.autessa.com/clients/aaf/inspect
 
{
  "policyName": "hr-portal",
  "sessionId": "chat-7f3a2c",
  "subjectId": "emp-20931",
  "phase": "TOOL_CALL",
  "toolName": "get_payroll",
  "args": { "period": "2026-10" },
  "toolGated": true
}

ResponseSAMPLE

200 OK
 
{
  "action": "CHALLENGE",
  "enforced": true,
  "reasons": ["AUTH_GATE"],
  "challenge": {
    "challengeId": "ch_9k2…",
    "methods": ["WORD_OF_DAY", "TOTP"],
    "prompt": "Please say: river, today’s word, seven."
  }
}

The problem

Your team may already run its own chat app, agent framework or voice stack. It still needs injection detection, verification, voice checks and limits.

How ai application firewall api works

STEP 1

Create a client API key

Authenticate with the Authorization header. Fetching the word of the day needs the Word of the day permission on the key.

STEP 2

Send each phase

USER_INPUT before your model reads the message, TOOL_CALL before a tool runs, TOOL_RESULT after it returns, FINAL_RESPONSE before the reply, VOICE for each piece of speech.

STEP 3

Keep a session and a subject

sessionId is your conversation ID; subjectId is your stable, opaque user ID that anomaly detection and daily limits learn from.

STEP 4

Act on the result

In Monitor, enforced is false so you can log; in Enforce, apply the action. For CHALLENGE, show the prompt and call challenge/verify.

Every setting

inspect

Check one phase of a turn against a policy.

voice/score

Score a clip for synthetic speech and the watermark.

challenge/start and challenge/verify

Run a verification.

authenticators

Enroll, confirm, read status and remove TOTP for a subject.

word-of-day

Fetch today’s word for your own staff app.

alerts, sensitive-record counts, dashboard

Read what the firewall recorded.

simulator

Run attacks against a named policy and read results.

Named policies and overrides

policyName picks a policy; overrides change sections for one call.

Who can do what

Developers

Call the API with a client key.

Admins

Create named API policies on the Settings page.

What to know

Pricing is the same for the API and for agents, charged only for checks that actually run.
If a usage limit blocks new work, inspect and voice/score are refused before anything runs and aren’t charged.

AI Application Firewall API: questions

Is there an API for LLM guardrails?

Yes. The Autessa AI Application Firewall API inspects each phase of a conversation turn and returns an action, reasons and findings.

Is it a proxy in front of my model?

No. It’s a turn-inspection API: your app calls it at each phase and decides what to do with the answer, so it works with any model or framework.

Turn on ai application firewall api.

Your own applications call POST /clients/aaf/inspect for each phase of a turn and get back an action with reasons, plus endpoints for audio scoring, challenges, authenticators and alerts. Start in Monitor and see what it would catch on your real traffic.

Turn on the firewall

Starts in Monitor · nothing changes for your customers until you enforce

100% customizable, code and all

Make Autessa AI Agent Firewall fully your own.

Use it out of the box, then change anything. Ask Autobot, Autessa’s AI builder, for a new field, screen, report, workflow or agent and it builds it into your copy of the app. Or open the code and edit it yourself. Nothing is locked: the app, its data and its code are yours.

Start free and make it yours