Agent Trust Manager Overview & Setup

Agent Trust Manager is the Titan product for detecting, classifying, and governing AI agent traffic on your platform. It identifies AI agents — from LLM crawlers to autonomous browsers — hitting your protected flows, tells you who they are, and gives you the controls to decide what happens to them.

What it does

Detects AI agents — Flags sessions as agentic based on AI-category user-agent matching and agent telltale signals.
Verifies agent identity — Where the agent supports Web Bot Auth (WBA), cryptographically verifies its signed request against the agent's published key, distinguishing a verified operator from a spoofed one.
Surfaces agent attribution — Every flagged session returns the agent's name and WBA verification status, so you know not just that traffic is agentic but which agent sent it.
Enforces policy per flow — Customer Managed Rules let you Allow or Block traffic based on agent identity, WBA status, risk score, and country, scoped per public key. A Monitor mode is available for baseline observation before enforcing.

How it fits with Titan

Agent Trust Manager sits on top of the existing Titan stack rather than replacing it — it doesn't require a new client integration. Your existing Arkose placement continues to collect signals as it does today; Agent Trust Manager offering adds agent classification and identity layer on top, surfaced through the same Verify API response and Real-Time Logs your integration already reads.

Sample Payload

 "agent_trust": {
        "detected": true,
        "detection_source": "web_bot_auth",
        "agent": {
            "name": "ClaudeBot"
        },
        "web_bot_auth": {
            "provided": true,
            "agent": "https://www.anthropic.com/",
            "key_id": "NzbLsXh8uDCcd-6MNwXF4W_7noWXFZAfHkxZsRGC9Xs",
            "signature_verified": true,
            "signature_fail_reason": null
        }
    }

For more details on the payload, please click here.

How to enable

Reach out to your Customer Success manager to get started.