Who we are, and what is actually assured.
We sell evidence, so this page states what exists today and what does not. Anything not listed as in place is not in place.
Reporting a vulnerability
Send findings to info@checkii.ai. Machine-readable contact details are at /.well-known/security.txt. We acknowledge within two working days and will agree disclosure timing with you. We do not pursue researchers acting in good faith within that contact.
The company
- Legal entity: MCR Technologies Limited, a private limited company registered in England and Wales, incorporated 18 November 2019.
- Company number: 12319980 (Companies House record)
- Correspondence: info@checkii.ai
- Data protection contact: info@checkii.ai
- Trading as: checkii.ai
Assurance status
| Item | Status |
|---|---|
| Cyber Essentials Plus | Planned |
| ISO/IEC 27001 | Planned, after Cyber Essentials Plus |
| Independent penetration test of checkii.ai itself, summary published | Planned |
| Data processing agreement and sub-processor list | In preparation |
| Sample engagement report and verifiable evidence bundle | Published |
| Account-free evidence verifier in the browser, for issued report bundles, sealed exports and DSSE-sealed governed-OSINT transcripts, with their transparency-log entries checked offline, and a published registry of the keys that sign evidence and of the log that records it | Published, 15 October 2026, transcripts 16 October 2026; the registry lists the sample key only until the hosted report-signing and transcript keys are published |
| Hosted beta: offensive and defensive planes isolated at the network layer (Kubernetes NetworkPolicy, governed egress proxy) | In place, verified 6 October 2026 |
| Hosted beta: signed deployment-acceptance record (backup restore, database and object-store failover, egress enforcement, worker containment, tenant isolation, tenant key recovery, alert delivery, capacity limits), maker and independent checker | In place, 6 October 2026 |
| Hosted beta: per-tenant key rotation on AWS KMS, activated by an administrator other than the one who created the version | In place, verified 6 October 2026 |
| Hosted beta: self-service password reset by email, sent from our own domain with DKIM, SPF and DMARC, bounces and complaints routed to the operations team | In place, verified 14 October 2026 |
| Hosted beta: the console's Audit & Assurance, privacy-records and trust-assurance sections enabled for tenants; trust assurance as point-in-time attestation, with customer shares not yet exercised on the hosted cluster | In place, 12 October 2026 |
| Multi-factor authentication on the console | Not yet; single sign-on through your identity provider (SAML, OIDC, SCIM) is the mitigation available today |
| Self-serve account creation | Not yet; after payment an operator provisions the workspace within two working days |
| Terms of service and privacy policy | With counsel; see /terms/ and /privacy/ |
We have not claimed a certification we do not hold, and we will not. When one is granted the certificate reference appears here with its scope and expiry.
How the platform handles your data
- Self-hosted: assessments, findings and evidence stay in your own estate. We receive nothing from a self-hosted deployment.
- Hosted: data is processed in the United Kingdom and the European Union. Evidence is encrypted at rest with a per-tenant envelope key, and tenant isolation is enforced in the database rather than in application code alone. Scanning workers reach targets only through a governed egress proxy, and the offensive plane cannot reach the defensive plane at the network layer: both are enforced by network policy on the hosted cluster and verified by probes from inside the live workloads, not only designed that way.
- Testing scope: an assessment runs only against assets whose ownership has been verified, under an approval recorded separately from the request. A job outside that boundary does not run.
Air-gapped deployments
"Air-gapped" here means no outbound network path from the assessment plane. Two things still need a deliberate, operator-controlled channel: vulnerability and signature feeds, and any model relay used by the governance module. Both are optional, both are logged, and neither is enabled by default.
This page was last reviewed on 14 October 2026 and is updated whenever a row changes state. Corrections to info@checkii.ai.