Agent identity verification
Check whether a request comes from the agent it claims to be. A familiar name in the user-agent header is a claim, not proof.
Kairal checks who an agent is, what it is trying to do and whether that activity belongs on your website. Give useful automation the access you choose.
/products/An agent reading your catalogue, collecting your articles or acting for a shopper needs different access. Kairal brings identity checks, purpose classification and access policies together so you can make those decisions in one place.
Check whether a request comes from the agent it claims to be. A familiar name in the user-agent header is a claim, not proof.
Separate search, training and shopping assistants. Review what an agent declares alongside the pages it visits and the requests it makes.
Choose which pages and actions each agent can access. Permission to read a product page does not grant permission to sign in or place an order.
An attacker can copy a recognised crawler’s name. Kairal checks the connection behind that claim and examines its requests, helping you distinguish genuine agents from impersonators.
IP 192.0.2.10IP 198.51.100.24Same claimed name. Different verification result.
Search helps people find your pages. Training collects material for models. Shopping assistants retrieve information or act for a customer. Treat each purpose separately, and keep uncertain traffic under review.
/products//categories//articles/01/articles/02/products/01/delivery/A request sequence alone does not prove purpose.
Let an agent read public information while protecting account pages and checkout. Set access policies around the pages and actions you choose to make available.
/products/01/availability//account/settings/Give shopping assistants access to product details and availability. Review repeated catalogue collection separately from requests that support a customer’s purchase.
Set different policies for search indexing and training collection. Keep decisions about how readers find your work separate from decisions about who can collect it.
Decide which listing pages agents can read. Apply separate controls to contact details, seller profiles and messaging.
Distinguish requests for event information from repeated checks of seat availability. Keep booking and purchase actions under their own access rules.
An identified agent still needs permission to act. Keep login, account changes and order submission subject to your authentication and purchase controls.
Define which endpoints agents can access and for what purpose. An agent’s identity does not replace the credentials or permissions your API requires.
Measure AI requests by company and purpose. Separate user-triggered visits, AI search and training collection, then follow how each changes over time. See which agents keep returning and which pages they read most.
Select a purpose to isolate its requests in the chart.
Selected leading companies · total requests
Review the agent’s claimed identity, the IP, the pages requested and the response applied. Follow requests across a session to understand why traffic was let through or blocked.
GET /products/01200GET /availability/200GET /account/settings/403/account/settings/The identity is verified, but the policy does not give this agent access to account settings.
Use that record to adjust access when a policy interrupts useful automation or leaves unwanted collection unchecked.
Start with the pages agents already visit and the actions you want to control. We’ll review your edge or server setup with you and agree where identity checks and access decisions belong.
Explore integrations| What you check | What it tells you | What still needs checking |
|---|---|---|
| User-agent header | The identity a request claims | Whether it genuinely comes from that agent |
| Verified identity | Evidence of who sent the request | Whether the requested action is permitted |
| Declared purpose | What the operator says the agent does | Whether its requests match that purpose |
| Pages and requests | What the agent accesses on your site | Whether that access fits your policy |
Knowing who sent a request helps you assess it. It does not establish permission to access every page or act for a customer.
Our founders built bots. Our team comes from offensive security, scraping and traffic analysis. We understand how automation is built, how it presents itself and how operators change it when access is blocked.
That experience shapes how we examine agent traffic: check the identity, follow the requests and assess whether the action belongs on your website.
About KairalTell us which agents you want to welcome, which pages they should reach and which actions need tighter control.
Discuss agent accessAgent Traffic Intelligence is in early access.