Severities and actions
Last modified: 25 July 2026
The three tiers
Rules carry a severity, and each severity carries an action. The shipped defaults:
| Severity | What lands there | Default action |
|---|---|---|
| 1 | Mild toxicity — insults, low-grade rudeness | warn |
| 2 | Harassment and hate — targeted abuse | mute, 300 s |
| 3 | Slurs and credible threats | mute, 3600 s |
actions:
severity-1:
type: warn
severity-2:
type: mute
duration-seconds: 300
severity-3:
type: mute
duration-seconds: 3600
warn and mute
type accepts warn or mute. Anything else is rejected at
startup rather than quietly ignored.
- warn — a visible chat warning to the player. No
duration-secondsis allowed on a warn; the plugin refuses to start if one is present. - mute — a timed voice mute.
duration-secondsis required and must be between 1 and 86400 (24 hours).
Muting severity 1 gets noisy quickly. The default is a warning for a reason.
What staff see
A flag alerts anyone holding voicesniffer.staff with the player, the severity and
the transcript. If a severity has no action configured, the flag is alert-only: staff are told, the
player is not touched.
Manual control is /vs mute and /vs unmute —
see commands and permissions.
Accuracy, honestly
Transcription is not perfect and the plugin does not pretend otherwise. Version one matches patterns — slurs and fixed harassment phrases — which is exactly the shape a transcriber gets right most reliably. General sentence accuracy is not what decides whether a flag was correct.
Two settings do most of the work in keeping false positives down, and both are on the capture page: the minimum utterance length, which drops clicks and breaths before they can be transcribed into something that looks like words, and the hangover, which keeps a sentence in one piece so a phrase rule can see all of it.