Quarterly Cloudinary cleanup console for storage and costs

Build me a media library cleanup console for Cloudinary that I open at the start of every quarter to get our storage and costs back under control. The audience is the ops or web lead who owns the Cloudinary bill, so this should read as a working surface rather than a report: I want to see what is bloating the account and act on it from the same screen.

Across the top, show a plan health strip built from Get Account Usage. Display storage, bandwidth and transformations, each as the amount used against our plan limit with the percentage used called out as the headline number. Make anything above 80 percent visually obvious. Credits cover storage, bandwidth and transformations together, and it is common to hit the transformation quota long before storage, so all three deserve equal weight in the strip.

Below the strip, show three triage queues, each built from Search Assets with its own query expression. The first is biggest files, sorted by size descending. The second is assets that have no tags. The third is assets not touched since a date I choose, where the stale threshold defaults to twelve months back from today and has a control to change it. Each queue loads independently and shows its own result count.

Every queue is a checkbox list. Each row shows a thumbnail of the asset, its file size, the folder it lives in, and its upload date. Give each queue a select-all control and keep a visible count of how many rows I currently have selected.

Once rows are selected I can act on them. Tag them as archive candidates with Manage Tags. Correct a bad name on a single row with Rename Asset. Write a keep reason onto an asset with Update Resource by Asset ID, saving the reason as contextual metadata on the asset itself so it travels with the asset outside this app. Remove the selected rows with Delete Resources by Public ID. Separately, offer a batch option that clears a whole retired campaign in one go through Delete Resources by Tag, where I pick the tag from the tags already present on the loaded assets or type one in.

Keep decisions must persist in the app, per asset. When I mark an asset as keep, remember that so the asset stops resurfacing in the next quarterly review. Show me how many assets are currently suppressed and give me a way to clear that list when I want a clean sweep.

Deletion needs guardrails. Before any delete runs, show a preview of exactly what is about to be removed: the list of assets, the count, and the total megabytes involved. Then require me to type a confirmation before it executes. After the delete completes, show the megabytes reclaimed and keep a running total of reclaimed storage for the session so I can see the payoff of the cleanup.

Two practical notes for the handlers. Cloudinary's admin endpoints page with a next_cursor, so respect it when loading queues instead of showing only the first page. The admin API is also capped per hour, so batch the bulk tag and delete calls rather than firing one request per asset, and surface a clear message if we hit the cap instead of failing silently.