POLICY BENCH / NAC–PB1 / DEFENSIVE OPERATIONS
Write the safeguards before enforcement
A credible policy says when a signal matters, who reviews it, what disruption is acceptable, and how a player can challenge a decision. Set those boundaries here. The template is generated in this browser and sent nowhere.
LOCAL POLICY INSTRUMENT / NAC–PB1
Build a defensive policy
Four operator choices. Five visible layers. No account, upload, or remote request.
GENERATED LOCALLY / 2026-07-18
Defensive enforcement policy
1. Scope
This policy covers player-behavior enforcement. Infrastructure protection, chat moderation, and world-visibility controls are documented separately so none are implied by the word anticheat.
2. Signals
Require corroboration before a signal changes a player's access. Require two independent signal families, or one signal plus direct staff observation, before action.
3. Action ladder
Use a reversible restriction only after corroboration; permanent action requires a completed review. Route durable actions to the shared review queue rather than relying on a single live alert.
4. Player safeguards
Controls begin in observation mode. A control that repeatedly burdens known-clean play is paused for review. Every notice names the affected policy and the review route.
5. Evidence and appeal
Record which independent observations agreed and which relevant observations did not. A second team member reviews permanent actions when available; any single-reviewer exception is recorded. Players may request review without admitting wrongdoing.
METHOD REGISTER / PUBLISHED DEFINITIONS
Evidence stays separate from policy
The builder documents operator choices. It does not claim those choices work. Measured claims require a named method, a versioned known-clean fixture, an untreated control where applicable, and an attached evidence record.
clean-play-signals / 2026-07-18
Clean-play signal observation
Count categorized detection events during a versioned known-clean fixture and compare distributions with an untreated control.
No measured result publishedresource-cost / 2026-07-18
Resource-cost comparison
Compare MSPT, CPU, heap, and garbage-collection distributions under identical versioned load.
No measured result publishedstack-compatibility / 2026-07-18
Stack compatibility record
Record whether a named server, proxy, translation, and client-mod stack completes a versioned fixture without an attributable conflict.
No measured result published6 known-clean fixture definitions are published for inspection.Read the fixture register orcompare fixtures with methods.