How Lockryn works
The deterministic path — rules and policies — is separate from AI. AI explains and proposes; only the response module changes the state of systems, after a policy check.
Diagram
Diagram
From an agent on your server to a person who sees the incident and can undo the action.
Step by step
Collection on the server
The log agent reads system, sign-in, web server, Docker and application logs. On the server it masks passwords, tokens, card numbers and personal data, applies the service's sensitivity profile and buffers to disk when the link is down. Separate agents watch file integrity, vulnerable packages and server resources.
Delivery to the core
Agents connect to the core themselves — outbound connections with mutual TLS only. Server service ports are not exposed to the internet; metrics and administrative access go through a private VPN channel.
Normalisation
The ingest module parses the sources, maps events to a common schema, adds geolocation, source network and service context from the registry, and writes the result to the event store tagged with the tenant.
Detection
Detection rules, together with time-window correlations, check new events on a schedule. Alerts from the agents on your servers and availability alerts go through the same code, and duplicates are merged.
Incidents and AI triage
Detections are grouped into incidents with a rule-based priority. AI then receives de-identified context (IP_1, USER_2 instead of real values) and answers strictly to a schema: verdict, priority, explanation, links to evidence and catalogue actions only. An invalid answer sends the incident to a person, marked “needs a human”.
Policies and response
A policy engine checks the proposed action: is it in the catalogue, is the target yours and not on the allow list, which autonomy level is allowed, is the rate limit exceeded. An allowed block is applied by the agent on your server and lifts itself when its time runs out (TTL).
Notifications and console
People get a short message in their language: what happened, where, how critical, what has been done and what is proposed. Details stay in the console; internal addresses and log fragments never go to messengers.
Knowledge base
On a schedule Lockryn pulls CVE/NVD, CISA KEV, EPSS, OSV, MITRE ATT&CK and CERT-UA and matches them with the software on your servers. You are notified only about what concerns you.
Autonomy levels
Every action has a default level and a ceiling it cannot be raised above.
| Level | Name | What happens |
|---|---|---|
| L0 | Observe | the event is only recorded |
| L1 | Notify | a person gets a message with a recommendation |
| L2 | Automatic | a reversible, low-impact action runs at once and can be undone in one click |
| L3 | With approval | the action runs only after a person confirms it |
| L4 | Forbidden | the system has no technical means to do it |
In the first release: notifications (L1) and temporary IP blocks (L2). For services with a medical, card or financial profile a block is L3; until approval arrives in the console, that means a recommendation and a notification.
What Lockryn does not do
- it does not hack back or scan other people's systems;
- it does not change your services' code — AI only proposes fixes;
- it does not replace antivirus on workstations;
- it does not send internal addresses, log fragments or secrets to messengers.
Want to try Lockryn on your services? Get early access