Responsible disclosure.

Email security@vortalis.ai with enough detail for us to reproduce the issue. A good report names the affected endpoint or component, the steps to reproduce, the impact you believe the issue has, and any proof-of-concept material. If you would prefer to encrypt your report, ask us in a first message with no sensitive detail and we will share a current public key. The machine-readable version of this contact information is published at https://vortalis.ai/.well-known/security.txt under RFC 9116.

The Vortalis production system is in scope: the governance proxy and its API surface, the public marketing site at vortalis.ai, and the admin dashboard. The public JWKS endpoint, the audit-verification surface, the tokenisation and unwrap paths, the policy-evaluation hot path, and the tenant-isolation boundary are the areas where a finding has the highest impact, and where we are most grateful for careful work. Authentication and authorisation flaws, tenant-isolation breaks, cryptographic weaknesses in the audit chain or attestation signing, and injection of any kind are all firmly in scope.

We ask researchers not to spend time on the following, because a report on them is unlikely to lead to a fix. Findings that require a compromised end-user device or a malicious browser extension. Volumetric denial-of-service testing against production, which we ask you never to perform. Reports generated solely by an automated scanner with no demonstrated impact. Missing security headers or cookie flags with no exploitable consequence. Social-engineering of Vortalis staff or customers. Issues in third-party services that Vortalis depends on but does not operate, which should be reported to those providers directly. The honest limits of what the platform itself defends against are documented separately at /product/security-architecture and are not vulnerabilities.

We will not pursue or support legal action against a researcher who acts in good faith under this policy. Good faith means you make a genuine effort to avoid privacy violations, data destruction, and interruption of our service; you only interact with accounts you own or have explicit permission to test; you do not exfiltrate, retain, or share any data you encounter beyond the minimum needed to demonstrate the issue; and you give us a reasonable opportunity to remediate before any public disclosure. If you are unsure whether a specific action is authorised, ask us first. We would much rather answer a question than receive a report from a researcher who guessed wrong about the boundary.

We commit to acknowledge your report within three working days. We commit to provide a substantive triage assessment, including our view of the severity and our intended remediation path, within ten working days. We commit to keep you informed at reasonable intervals while we work on a fix, and to tell you when the fix has shipped. We commit to credit you in our acknowledgments page at /security/acknowledgments unless you ask us not to. Vortalis does not currently operate a paid bug-bounty programme; this is a coordinated-disclosure policy, and the recognition we offer is public credit and our genuine thanks.

We follow a coordinated-disclosure model. The default embargo period is ninety days from the date we acknowledge the report, which gives us time to ship and deploy a fix and to notify affected customers where their contracts require it. If a fix is ready sooner, we are happy to coordinate an earlier disclosure with you. If a fix is taking longer because the underlying change is structural, we will tell you why and agree a revised date with you rather than letting the embargo lapse silently. We will never ask for an indefinite embargo. If we cannot fix an issue, we will say so plainly and document it on the limitations page rather than leaving you waiting.

Vortalis is the trust layer for AI agents. The whole product is a claim that an independent party can verify what an agent did, that sensitive data is shielded by default, and that authority can be revoked in seconds. A security researcher who finds a hole in any of those claims is doing the most valuable work anyone can do on this platform, because the alternative is that a customer or a regulator finds it first. We treat disclosure reports as a gift, not a nuisance. The cryptographic audit chain exists precisely so that an outsider can check our work; a vulnerability report is the same instinct applied to the rest of the system.