T3 Code Partitions
T3 Code keeps per-partition HTTP caches under its Partitions directories so each preview or embedded web context can reuse downloaded responses, scripts, stylesheets, images, and Chromium network cache entries instead of fetching them again. Over time, old preview-partition cache data accumulates across both the legacy t3code profile and v2 profiles, and stale cached responses can keep serving outdated web assets inside previews. Kudu removes only preview-partition HTTP caches older than one day from %AppData%/t3code/Partitions and %AppData%/t3code-v2/Partitions, while preserving partition sessions, storage, and workspace data.
Why clean T3 Code Partitions?
- A stale cached HTML, JavaScript, or CSS response in a preview partition can keep showing an older build, so edits appear missing even though the source files were saved correctly
- Expired Chromium network cache entries sometimes survive longer than expected and cause embedded previews to load broken styling, missing icons, or partial pages until the cache is cleared
- Preview partitions created across multiple projects and v2 profiles leave behind old disk cache blocks, so %AppData% grows noticeably and roaming profile sync or backups take longer
- If a local dev server changes asset hashes or cache headers, the partition may keep reusing outdated responses, which shows up as mismatched UI behavior between the browser and T3 Code preview
- Corrupted cache records inside a partition can cause repeated preview reloads, blank embedded pages, or requests that hang until T3 Code refetches everything cleanly
- Old HTTP cache data from no-longer-used preview partitions wastes disk space without helping current work, especially when switching between many branches, demos, or temporary workspaces
Cache paths Kudu targets
Windows
%AppData%/t3code/Partitions |
%AppData%/t3code-v2/Partitions |
Common questions about T3 Code Partitions
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.