Admin Guide#

Common day-to-day scenarios and the command/config combination that handles each.

Silencing a false positive for one player#

/ff exempt <player> check:<id> 30m "reason" (player must be online). Use category:MOVEMENT or a wider scope only when several related checks are all firing on the same legitimate behavior. See the worked example in API Integration for the same pattern driven from your own plugin instead of a staff member's hand.

A check is too sensitive server-wide#

Edit its parameters in checks.yml — see the Check Parameter Reference for every check's actual keys and the Tuning Guide for how to use them safely — then /ff reload, no restart needed. Invalid YAML or an out-of-range value disables only that one check with a logged reason; the rest of the snapshot still loads.

A check keeps producing findings you disagree with, permanently#

/ff check <id> disable. This is instant and doesn't require editing a file, but doesn't survive a /ff reload/restart unless you also update checks.yml — the file is the durable source of truth.

A check is crashing#

FlagForge automatically disables any check after 5 consecutive internal exceptions and logs it at SEVERE (/ff checks shows DISABLED (repeated internal errors) next to it) — this is a bug, please report it. /ff reload or re-enabling the check gives it a fresh start once the underlying issue is fixed.

Emergency stop#

/ff enforcement off "reason" disables every preventable response server-wide immediately, without touching detection/alerting/checks.yml. Detection keeps running so you don't lose evidence while investigating.

Clearing a bad decision#

/ff vl <player> clear check:<id> "reason" (or category:<name> or all) resets accumulated violation level without touching stored history.

Reviewing what actually happened to a player#

/ff inspect <player> for their live profile (game mode, world, active exemptions, current violation state), /ff history <player> for their persisted finding history, and /ff finding <id> for one finding's full structured evidence. For direct database access instead — building an external dashboard, or an incident spanning many players — see the Data Model reference.

An administrator suspects item duplication#

Don't chase it manually per-player: actions.yml's confirmed-inventory-generation correlation rule (see Configuration § actions.yml) already matches the evidence-family combination that indicates dupe activity, and suspected-dupe-backup-trigger can fire a backup snapshot automatically on much weaker evidence — see Tuning Guide § Triggering a backup on suspected dupe evidence. Run /ff correlations to check both rules' current reachability given your coverage-complete setting before assuming either is silently doing nothing.