Apps
Google Cloud CLI Cache
Google Cloud CLI writes rotating command logs and stores cached API responses, discovery documents, credential metadata, and other HTTP cache files under its gcloud configuration directory to avoid re-fetching service information on every command. Over time those log files and cached API artifacts can become stale or accumulate across many projects and accounts, leading to confusing command behavior and wasted disk space. Kudu removes the gcloud logs directory and cached API data while leaving your configured accounts, active project settings, and credentials untouched.
Why clean Google Cloud CLI Cache?
- Stale cached discovery documents can make newer service fields or flags appear missing, so commands fail with unrecognized argument or invalid field errors until metadata is refreshed
- Cached API responses from an older project or account context can surface outdated resource listings, making instances, buckets, or IAM changes seem not to exist
- Accumulated gcloud log files grow quietly after repeated auth, deployment, and scripting runs, consuming space in the user profile without affecting saved configurations
- Corrupted HTTP cache entries can trigger repeated command retries or unexpected parsing errors, which users notice as commands hanging longer than usual before failing
- After API changes or organization policy updates, old cached service metadata can cause misleading validation errors even though the same command works once the cache is rebuilt
- Large log directories make troubleshooting harder because fresh errors are buried under months of old command output, especially on systems running scheduled gcloud jobs
What gets cleaned
Cache paths Kudu targets
Windows
%AppData%/gcloud/logs |
%AppData%/gcloud/cache |
macOS
~/.config/gcloud/logs |
~/.config/gcloud/cache |
Linux
~/.config/gcloud/logs |
~/.config/gcloud/cache |
Frequently asked
Common questions about Google Cloud CLI Cache
Related cleaners
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.