# Anti-Cheat Architecture

The anti-cheat subsystem is split into client telemetry evaluation and
server-authoritative validation.

## Client Side

Client checks evaluate supplied telemetry snapshots. The framework covers memory
integrity, speed drift, aimbot heuristics, graphics hooks, movement anomalies,
process snapshots, driver integrity, signed heartbeats, source obfuscation,
evidence capture requests, signature updates, salted hardware fingerprints,
overlay hooks, enforcement modes, and minimal-footprint policy.

The client module does not enumerate operating-system resources directly. Native
or platform-specific collectors should provide snapshots to the pure detection
functions.

## Server Side

Server checks are authoritative and should gate competitive gameplay:

- Replay movement inputs and compare the result with reported positions.
- Rewind target history for lag-compensated hit validation.
- Validate damage against weapon damage, range, falloff, and fire rate.
- Reject unauthorized resource and economy gains.
- Flag population outliers using configurable statistical baselines.
- Open review cases with replay frames and structured violation evidence.
- Apply temporary, permanent, hardware, IP-range, or shadow bans.
- Track appeals and reviewer resolution notes.
- Collect player reports and trusted reporter status.
- Emit analytics data for operations dashboards.
- Submit evidence to third-party anti-cheat providers through adapters.
- Run deterministic simulations for known cheat classes.

## Operations

Ranked modes should run full client checks plus server validation. Casual modes
can use minimal-footprint client checks while retaining server authority.
Single-player modes can opt out of enforcement unless a game mode explicitly
requires integrity checks.

Evidence capture should be triggered only for high-severity findings. Hardware
fingerprints are salted and should be stored only for enforcement workflows with
clear retention policy.
