Lockryn

Security of the platform itself

Lockryn has access to telemetry from everything it protects. So we hold the platform to the same requirements as our customers' services — and stricter ones.

Principles

Nine rules the core is built on. Each one is about keeping a fault in one part from opening up the rest.

  • Tenant isolation from day one

    The tenant tag is in every record, bus message, event-store row, cache key and AI request. The database runs with forced row-level security (RLS), the event store through a layer that adds the tenant filter. Code without a tenant isolation test is not accepted.

  • Data in the EU

    The core, storage and backups are in data centres in Germany and Finland. The core runs in a separate account and location from the services it protects. Raw events and tenant data do not leave our infrastructure in the EU.

  • AI advises, policies decide

    The AI module has read-only tools. Any change of state goes through the response module and the policy engine. The policy fails closed: input outside the contract is a denial.

  • De-identification for AI

    The cloud model gets stable placeholders (IP_1, USER_2, HOST_3); the mapping table stays in the EU. Logs and headers are passed as data, not instructions; fragments that look like commands are flagged as suspicious.

  • Hash-chained audit log

    System and human actions are append-only; the database forbids changing or deleting records. Each tenant has its own chain, and a change to any record is caught by verification.

  • Sign-in and roles

    Console sign-in through a separate identity service with passkeys. Roles: owner, administrator, analyst, viewer; a viewer cannot turn on the kill switch or cancel actions. The provider admin is separate from the customer console.

  • Few open doors

    Only the console and log intake (TLS with agent certificates) are reachable from the internet. Administrative interfaces and SSH are reachable only through a private VPN channel and only with keys.

  • Limits and a kill switch

    Every action has an expiry or a rollback and starts as a dry run. A cap on automatic blocks per interval switches the tenant to “notify only” by itself; the kill switch does the same.

  • Secure development

    CI runs secret scanning, static analysis, vulnerable-dependency and licence checks. Secrets are stored encrypted, never in code. A STRIDE threat model is maintained.

What Lockryn sees

Collected

  • sign-in, web server, system and application logs — after masking;
  • resource and availability metrics;
  • file changes and the list of installed packages;
  • alerts from the agents and monitoring.

Not collected for medical, card and financial services

  • request and response bodies;
  • query strings;
  • cookies and the Authorization header;
  • passwords, tokens and card numbers are masked on the agent for every service.

Found a vulnerability?

Write to office@softsoledgeit.com with the subject “Lockryn — vulnerability”. Contacts for researchers are also in security.txt.

Want to try Lockryn on your services? Get early access