Solver

Solver 0.7.0

Sanctions, custom messages, and a better chat moderator. This release brings a real warn/mute/kick/ban system, every player-facing message is now yours to customize, and Fyrx's moderation calls now come with a confidence score instead of a flat yes/no.

Highlights

• New messages.yml — every player/staff-facing message (help, errors, sanctions, moderation alerts) is now configurable with full MiniMessage support, in your own language folder (lang/{locale}/messages.yml).
• New Sanctions system (BETA) — /solver warn|mute|tempmute|kick|ban|tempban <player> [duration] <reason>, plus /solver unban|unmute|unwarn, /solver history <player>, /solver check <id>, and /solver note <player> <text>. Every sanction gets a real ID, persists in sanctions.db, and shows its own correctly-worded message (a kick doesn't say "banned"). New and still under active testing.
• Bans are actually enforced now — reconnecting while banned is rejected at login, not just a one-time kick.
• Mutes always work — even with AI chat-moderation turned off, /solver mute now really blocks chat.
• New: Confidence Meter — every moderation verdict now shows how sure Fyrx is that it's correct (0-100%), separate from how severe the behavior is. Visible in staff alerts, /solver moderation test, and averaged in /solver moderation status.
• /solver moderation status now shows a countdown to the next automatic chat scan, so you're never left wondering if it's working or just waiting.
• New: PlaceholderAPI support (BETA) — %solver_muted%, %solver_warns%, %solver_active_sanctions%, and more, for your scoreboard/tab-list/chat plugins. Optional, does nothing if you don't have PlaceholderAPI installed.
• config.yml/messages.yml no longer go stale on update — new options added by a future update now merge in automatically instead of silently only working off an invisible default.
• AI Chat Moderation, out of beta — staff/admins can now be exempted so they're never muted by their own system, full localization in all 9 languages, and a fixed bug where a single word ("harmless"/"toxic") replaced a real per-category, per-offender verdict.

More details: docs.jeamcube.com/en/solver/changelog