Kyvvu

Security

Security & responsible disclosure.

Kyvvu builds the Agent Security Kernel for AI agents. We take the security of our products and systems seriously and want reporting a vulnerability to be easy and safe for you.

Kyvvu BV, Archipelstraat 13, 6524 LK Nijmegen, the Netherlands. Chamber of Commerce: 99594234.

Last updated: 4 November 2026

This page is for information only. It does not create any contractual obligation for Kyvvu BV. Kyvvu's commitments to customers are set out exclusively in the Kyvvu EULA and its Data Processing Agreement (Annex 1), the Kyvvu General Terms and Conditions, and, for self-service accounts, our Terms of Service.

Our security approach

  • Kyvvu operates an information security management system (ISMS) aligned with ISO/IEC 27001:2022 and is working towards certification. We will not claim to be certified until a certificate has been issued.
  • By default, Kyvvu runs inside the customer's own environment. Agent content and policy data stay on the customer's systems.
  • Where Kyvvu hosts components (the Cloud Control Plane), these are hosted in the EU: AWS Frankfurt (eu-central-1) as primary region, with disaster-recovery copies in AWS Ireland (eu-west-1). Content delivery and the web application firewall run on Amazon CloudFront's global edge network.
  • Personal data that Kyvvu processes on behalf of customers is governed by the Data Processing Agreement. The current list of sub-processors is available on request via privacy@kyvvu.com.
  • Customers are notified of personal data breaches and security incidents as set out in their agreement with Kyvvu.

How to submit a vulnerability

Email security@kyvvu.com. Our security.txt points to this policy and to the same mailbox.

Please include:

  • a description of the vulnerability and its impact;
  • step-by-step reproduction steps, with any proof-of-concept code, screenshots or videos;
  • the affected URL, endpoint or product and version, where applicable; and
  • your contact details, if you would like updates and credit.

Reports may be submitted in English or Dutch.

Scope

In scope:

  • Kyvvu's public website and marketing domains
  • The Kyvvu SaaS control plane and its public APIs
  • Kyvvu policy-execution components, where Kyvvu operates them

Out of scope:

  • Third-party services and SaaS products that Kyvvu uses but does not operate (please report these to the vendor)
  • Customer environments, including Kyvvu components run by a customer; testing these requires the customer's explicit prior consent
  • Social engineering of staff, contractors or customers, and physical attacks
  • Denial-of-service (DoS/DDoS) testing and automated scanning that generates excessive traffic
  • Missing best-practice headers or configuration without demonstrable, exploitable impact

Kyvvu will not take legal action against researchers who report vulnerabilities in good faith and in accordance with this policy, provided they stay within scope, do not access or modify data beyond what is strictly necessary to demonstrate the issue, and respect the disclosure timeline. Testing in accordance with this policy is considered authorised.

This safe harbour does not cover actions that violate applicable law or that affect data or systems of Kyvvu's customers or other third parties. Kyvvu may withdraw the safe harbour if these conditions are not met.

Terms and conditions

  • Do not publicly disclose a vulnerability until Kyvvu has had a reasonable opportunity to investigate and remediate it, typically 90 days from the date of your report.
  • Do not access, modify, exfiltrate or destroy data belonging to Kyvvu or its customers.
  • Do not run tests that disrupt the service for other users (no DoS, no high-rate automated scanning).
  • Do not test customer environments without the customer's explicit prior consent.
  • Reports must contain enough detail to reproduce the issue.

What you can expect from us

  • Acknowledgement of your report within 5 working days.
  • An assessment and target remediation timeline within 10 working days.
  • Status updates while we investigate and fix the issue.
  • Public credit on our security acknowledgements page, if you wish.

Kyvvu does not currently offer a paid bug bounty or any other monetary reward.

Acceptance criteria and prioritisation

Valid reports are triaged by Kyvvu and assigned a severity of Low, Medium, High or Critical, based on exploitability and impact on confidentiality, integrity and availability.

Reports that are typically rejected:

  • theoretical issues without demonstrable practical impact;
  • missing best-practice headers without a working exploit;
  • findings against out-of-scope or third-party systems; and
  • vulnerabilities already known to Kyvvu.

Terms of Service · Privacy Policy · Acceptable Use Policy