Throttle AI workloads, keep your PC smooth.
A one-tab MAUI control surface and local daemon that manages concurrency, rate limits, and thermal pressure for local AI — fewer stutters, fewer thermal spikes.
Start daemon
dotnet run --project src/LeaseGateLite.Daemon
Status
GET /status → { cpu: 42%, heat: "Calm", queue: 2 }
Throttle
POST /config → { maxConcurrency: 3, cooldown: 5s }
Features
Everything you need to keep AI workloads under control.
Live system metrics
Real CPU%, RAM%, queue depth, and heat state (Calm/Warm/Spicy) via Windows PerformanceCounter and GlobalMemoryStatusEx.
Adaptive throttling
Soft and hard thresholds with recovery rate and smoothing — automatically backs off when your system gets hot.
One-tab control panel
MAUI desktop app with an auditable checklist layout — every setting visible, every action explicit.
Per-app profiles
Override concurrency, rate limits, and presets per application — give your browser one budget and your batch jobs another.
Optional auth
Start with --require-auth to generate a local token and lock down the daemon API. Simulation endpoints are gated behind --enable-simulation.
195 tests
Config validation, flood simulation, redaction, diagnostics — all deterministic via FakeSystemMetrics dependency injection.
Usage
Start the daemon
# Start the local API daemon
dotnet run --project src/LeaseGateLite.Daemon
# Daemon listens on localhost:5177 Launch the control panel
# Build and run the MAUI app (Windows)
dotnet build src/LeaseGateLite.App \
-f net10.0-windows10.0.19041.0
dotnet run --project src/LeaseGateLite.App \
-f net10.0-windows10.0.19041.0 Daemon Endpoints
Local REST API on localhost:5177.
Presets
Three built-in profiles to match your hardware.
Quiet
For laptops — conservative limits, aggressive thermal protection, lower concurrency.
Balanced
The default — sensible limits for typical desktop use with moderate AI workloads.
Performance
For desktops with headroom — higher concurrency, relaxed thresholds, maximum throughput.