Documentation

SecAI documentation

What SecAI installs, collects, watches, blocks and changes, written from the implementation and dated. Start with the quickstart; the Trust Center holds the same facts with their sources.

Getting started

QuickstartFrom an account to a monitored server in about five minutes.InstallWhat the install script does, in the order it does it.Supported systemsWhere the agent runs, and what it recognises there.UninstallFrom the dashboard, or by hand in six commands.

The agent

The agentOne static binary, reporting every minute, verified by signature.Data collectedThe telemetry categories, from the agent's payload, and what never leaves the server.PermissionsWhy root, what it reads, and the short list of what it may change.Network requirementsOutbound HTTPS to one host. No inbound port.

Detection

Threat detectionWhat the agent watches for, by signal, and how findings are rated.File integrityThe paths, the hashes, and the two cases where text travels.CVE monitoringInstalled packages, matched against OSV.dev every twelve hours.AI auditsWeekly and on demand, evidence attached, within your plan's allowance.

Response

IP blockingThree sightings in thirty minutes, one firewall rule, gone in 24 hours.Automatic modeAdds exactly one thing: blocking attacking addresses without asking.Approval modeAwaiting Approval: everything waits, including a block.RollbackEvery configuration change keeps a copy first; undo is one click.

Integrations

APIA read-only data API for your own tools and AI agents, keyed per organisation.WebhooksA signed JSON POST for critical alerts.MSSP and resellersClient organisations under one console, with recorded access and your own brand.

Help

TroubleshootingThe usual causes, in the order to check them.

Reviewed against the implementation on 2026-09-20.

Trust Center