Block attackers. Save your CPU. See everything.

Fast, and light

Detection runs in memory and bans land in nftables the instant a threshold is crossed. The agent uses a fraction of one core even on a busy host, and a flood makes it more selective — never slower.

Security that pays for itself

The scrapers and request floods that eat your CPU are the same addresses that probe your logins. Banning them gives the processor back to real visitors: faster sites, fewer incidents.

Every decision shows its logs

Each ban carries the lines that convicted it, and the console shows who is approaching a threshold before they cross it. You tune a rule on evidence, not on guesswork.

A console that keeps up

Live feed, world map, one search box for an address, a rule, a server or a network; ban, lift or exempt in one click; a public unblock page where a blocked visitor checks their address and asks to be let back in.

Fleet-wide, on your terms

Choose which servers offer their bans to the others and which enforce what the fleet shares. An enrolled machine receives its rules and their updates — and keeps defending itself when the controller is unreachable.

Reports where it counts

Native reporting of offenders to AbuseIPDB, with an allow-list of what may leave: usernames and log lines never do. Our own base is next.

API first

Everything the console does is one call an API token can make — from a script, a hosting panel, or the shield command on your desk.

Open source, free, European

Agent and controller are AGPL-3.0: read the source, build it, run it without paying anyone. The company, the servers and the data are in France.

Agent source · Controller source

Friendly to humans and to AI agents

Configuration is a handful of readable files, commands explain themselves, nothing is hidden. An AI coding agent — Claude Code, for one — operates and extends it with ease.