Apps

Kudu

Kudu stores several Chromium and Electron runtime caches under its Windows profile, including HTTP disk cache entries in Cache_Data, V8 compiled script data in Code Cache, GPU raster and shader data in GPUCache, and rotating diagnostic log files in logs. These files speed up loading the app shell, web assets, and hardware-accelerated rendering, but they can become stale after app or graphics driver changes and grow larger than needed over time. Kudu removes only those temporary cache and log files while leaving your settings, accounts, saved preferences, and any active files protected by recency and lock checks.

Why clean Kudu?

  • Outdated HTTP disk cache entries in Cache_Data can make Kudu load an older web bundle, showing broken layouts, missing icons, or UI elements that do not match the current release until the cache is rebuilt
  • Invalidated V8 bytecode in Code Cache after a Kudu update can cause a noticeably slow first launch, longer window startup, or a brief white screen while scripts recompile
  • Stale GPUCache shader and raster data from a previous graphics driver version can trigger black panels, flickering, or choppy scrolling until Chromium regenerates the cache
  • Corrupted cache records can leave parts of the interface stuck loading, with spinners that never finish or views that only appear after signing out and back in
  • Accumulated log files in the logs folder can consume hundreds of megabytes over time, especially after repeated startup errors, leaving users wondering why Kudu's profile keeps growing
  • A bloated mix of web cache, code cache, and GPU cache can increase profile size and make troubleshooting harder because Kudu keeps reusing bad temporary data instead of fetching clean assets
What gets cleaned

Cache paths Kudu targets

Windows

%AppData%/kudu/Cache/Cache_Data
%AppData%/kudu/Code Cache
%AppData%/kudu/GPUCache
%AppData%/kudu/logs
Frequently asked

Common questions about Kudu

Free & open source

Download Kudu and reclaim your disk space.

Available on Windows, macOS, and Linux. No account required, no feature gates, no telemetry without consent. All cleaning targets are open source and community-auditable.