convex-file-storage

Call a mutation that returns ctx.storage.generateUploadUrl(), then POST the file to that URL with the correct Content-Type header. The response contains a storageId, which you save in your own table via a second mutation.

Use ctx.storage.getUrl(storageId) inside a query to resolve a storage ID into a servable URL. Return the URL alongside file metadata so React components can render images, embed PDFs, or offer download links.

Query the _storage system table with ctx.db.system.get(storageId) to retrieve contentType, sha256, size, and creation time. The older getMetadata approach is deprecated in favor of this system table access.

Yes, actions running in the Node environment can create a Blob from generated content and persist it with ctx.storage.store(blob), which returns a storage ID. This works for PDFs, AI-generated images, or any binary output.

The most common cause is missing or incorrect Content-Type headers during the POST to the upload URL. Also verify the storage ID is valid and that the file was not deleted from storage while its database reference remains.

Fetch the database record, call ctx.storage.delete(file.storageId) to remove the stored object, then delete the database row with ctx.db.delete. Skipping the storage deletion leaves orphaned files that consume storage quota.