Argos

Why it exists

A twenty-person company has the same security obligations as one with two thousand, with nobody dedicated to it. It falls to the IT person who also looks after the printers.

What is on the market is built for the other size: licensed per asset, deployed over weeks, and it needs somebody who knows how to run it. So that company ends up with an inventory spreadsheet nobody maintains, backups nobody has ever restored, and a folder of compliance documents written once.

The platform

A self-hosted platform that brings into one process what today lives in seven places. The asset inventory, vulnerabilities and threat intelligence. Compliance with the Spanish ENS framework. And backup status, the device fleet and staff training.

It is not a dashboard showing seven things. An asset discovered in the inventory turns up on its own in vulnerability management. A flaw found there knows which ENS control it affects, and that control knows what evidence it is missing. Joining it up works because the data travels, not because it fits on one screen.

Self-hosted because the customer it targets does not want their asset inventory in anybody’s cloud, and because installing it has to be an afternoon.

I built it with agents

A seven-module platform is something I build by directing AI agents, with the rules written as files: the architecture, the boundaries of each module and what each one must not touch.

The design and the judgement are mine, and so is verifying that what comes out works. The volume of code is moved by the agents. Talos came out of here with its 147 checks, which is the proof the method holds up for a whole module.

Modularity

It is built in modules and some of them stand on their own. Talos, the hardening audit, came out of here and I published it openly.

That forces the module not to depend on the platform database, its user system or its report format, and it costs more to design that way. In exchange Talos gets used standalone and gets real usage, which a locked-in module never would. And keeping the boundaries clean left the rest of the platform easier to change, which was not the goal but is how it turned out.

Why SQLite

It is the decision I get questioned about most. The target is one installation per company, with tens of machines and a handful of concurrent users. At that size, a database that is a file removes one service to install, another to update, another to back up and another that can fail to start on a Monday.

Backing up the whole platform becomes copying a file, and restoring is putting it back. In a product whose argument is that you do not need anybody to maintain it, that is not a shortcut, it is the thing being sold.

Status

In development, and the code is not public because it is a commercial product. What I do talk about is the architecture and the decisions, which is what this page holds.