Run a media library audit

Most media libraries grow without anyone watching. Full-size photos straight off a phone, a pile of screenshots, and a long tail of attachments with no caption or description. Before you clean any of it up, it helps to have the library describe itself: how big it is, what it is made of, and where the gaps are.

This is the orientation prompt for your media. It reads the library and hands you a briefing, and it cannot touch a thing.

Give me a read-only audit of my media library. Change nothing.

Up top I want the total item count and a breakdown by file type (jpeg, png,
gif, webp, and anything else). Then flag the items missing a caption or a
description, with a count of each. Then read the [15] largest items and list
their file size and dimensions so I can see what is bloating the library.

Lay it out as a short briefing I can skim in under a minute, and put a one-line
takeaway at the very top: the single thing most worth acting on, whether that
is a stack of oversized originals or a mass of items with no caption.

Why this is safe to run

Every ability behind this prompt is a read. Counting the library is a separate ability from listing it, and none of these can write, so the worst case is that a read you left switched off quietly returns nothing rather than changing anything. Reading the largest items pulls their size and dimensions without altering the files at all.

This is a natural first step before the “bulk-update media captions and descriptions” recipe: run the audit, see how many items are missing a caption, then decide whether the fill-in job is worth your time.

Already connected? Skip the paste. Ask your agent for the “media library audit” recipe, or just ask it to size up your uploads, and it runs the same reads on your own install.