v1.16.0

New

AI assistant integration: the server can now talk directly to AI assistants (such as Claude) through the Model Context Protocol (MCP). Point your assistant at this server and you can ask things like "how many attacks did I get this week?", "is this IP address blocked anywhere?" or "show me the latest decisions" in plain language, without using the API by hand.
IP lookup for assistants: your assistant can check whether any IPv4 or IPv6 address appears in your blocklists or allowlists and tell you which lists contain it.
Everything the assistant can see or do goes through the same rules as the apps: if you set an API password, the assistant needs it too; if you didn't, it just works.
Filter by date range: when listing attacks or blocking decisions, you can now ask for only what happened during a specific period — just a starting moment, just an ending moment, or both. Handy for answering questions like "what happened last weekend?" without scrolling through everything.
This works everywhere the lists are shown: through the API and also when you ask your AI assistant for your attack history.

New optional MCP_RATE_LIMIT setting: if you want to limit how often assistants (or anyone) can query the MCP endpoint, you can now set a separate limit for it, independent from the one that protects the app API. Left empty, there is no limit — exactly like before.
Cleaner console logs for the AI assistant endpoint: the API no longer prints one raw HTTP line per MCP request (they only said someone was connected, not what they did). Instead, every assistant action is now logged with the name of the tool it ran, e.g. [mcp] tool: 'get_alerts_stats', so the console shows what the assistant actually asked for.
Internal modernization: updated build tooling (pnpm 12, TypeScript 6, faster dev server). Nothing visible changes for you.
The whole server was reorganized into clearly separated layers (web interface, business rules, data access), which makes future features faster to build and easier to review. Everything works exactly as before: no endpoint, setting, or behavior has changed.
Quality guardrails: the server is now backed by a comprehensive automated test suite (more than 1,100 checks covering the web endpoints, the database layer, the scheduler, and the configuration), so regressions are caught automatically before they reach a release.

General statistics now always deliver their numbers under the correct field names for countries, organizations, scenarios, and targets, so the mobile apps can read and display them reliably.

Dates and times for the new date range filters are always read in the UTC time zone, so results are the same no matter where the server lives.
If you have been using the beta versions, this stable release includes everything from v1.16.0-beta.1 to v1.16.0-beta.4. No manual action is needed: just update the API and keep your current settings.
No configuration or stored data changes. Updating is safe and requires no manual action: just update the API and keep your current settings.