Media library
Images, video and files in local S3 storage.
3 fileslocal S3 storage
| title | name | description | url | ext | type | crop | size | dimensions | created | |
|---|---|---|---|---|---|---|---|---|---|---|
| apple | seed-apple.png | — | http://localhost:3300/media/df6a428c-f804-4ba0-ae65-5ba2a9219819/file | .png | image/png | — | 22.4 KB | 1024×1024 | 2026-10-11 13:50 | |
| orange | seed-orange.png | — | http://localhost:3300/media/06f8df3e-1532-4791-a27b-2699ca2c2fa6/file | .png | image/png | — | 22.6 KB | 1024×1024 | 2026-10-11 13:50 | |
| development-loop-2026.jpg | development-loop-2026.jpg | — | http://localhost:3300/media/be3d6a24-686e-4378-82a7-04ce37173d07/file | .jpg | image/jpeg | — | 83.4 KB | 1200×780 | 2026-10-11 13:50 |
Object storage — files, compared with the database
What it holds. The files themselves: images, video, audio, documents. Bytes, not rows. It streams them, serves them with cache headers, crops images and trims video.
How it differs from the database. The database holds facts about a file — its name, size, type, duration, who uploaded it and when. This holds the file. The split is not bureaucracy: files inside a database bloat it and make every backup slow, while facts kept only in a filesystem cannot be searched, filtered or joined. Each keeps what it is good at, and they point at each other.
What it costs. Disk, and nothing else. No model, no per-request billing — the storage lives on your server, so a file served a million times costs the same as one served once.
Where it is weak. It knows nothing about content. A PDF is a sequence of bytes to it; until something extracts the text and puts it in vector memory or the knowledge base, nobody can ask a question about what is inside.