Attachments

Split down the middle: v2 reads attachments, v1 uploads them. v2 has no create endpoint at all, which is one of the clearest reasons v1 still ships.

import { createV1Client, createV2Client } from 'confluence.js';

const v2 = createV2Client({ host, auth });
const v1 = createV1Client({ host, auth });

Upload

await v1.contentAttachments.createAttachment({
  id: '12345',
  attachments: { filename: 'report.pdf', content: bytes },
});

content takes whatever you have:

Type	Use
string	text you already have in memory
Buffer / Uint8Array	the usual for a file read into memory
Blob / File	browser input, or fs.openAsBlob()
Readable (Node stream)	a large file you do not want to buffer
ReadableStream	a fetch response body, piped straight through
AsyncIterable<Uint8Array>	anything else that yields bytes

import { readFile } from 'node:fs/promises';

await v1.contentAttachments.createAttachment({
  id: '12345',
  attachments: { filename: 'report.pdf', content: await readFile('report.pdf') },
});

Several at once — pass an array:

const created = await v1.contentAttachments.createAttachment({
  id: '12345',
  attachments: [
    { filename: 'notes.txt', content: 'first' },
    { filename: 'data.csv', content: 'a,b\n1,2' },
  ],
});

console.log(created.results.map(result => result.title));

No FormData, no header

2.x made you build the FormData and remember X-Atlassian-Token: no-check. Neither is your job now — the call encodes the multipart body and sends the XSRF header itself.

Upsert

createAttachment (POST) adds; createOrUpdateAttachments (PUT) adds or versions an existing one with the same filename:

// First call creates it, the second adds a version — same id both times.
const first = await v1.contentAttachments.createOrUpdateAttachments({
  id: '12345',
  attachments: { filename: 'report.pdf', content: v1Bytes },
});

const second = await v1.contentAttachments.createOrUpdateAttachments({
  id: '12345',
  attachments: { filename: 'report.pdf', content: v2Bytes },
});

second.results[0].id === first.results[0].id; // true

Replace the data of one attachment

await v1.contentAttachments.updateAttachmentData({
  id: '12345',
  attachmentId: 'att67890',
  attachment: { filename: 'report.pdf', content: newBytes },
});

Rename, or edit properties

const attachment = (await v2.attachment.getAttachmentById({ id: 'att67890' }));

await v1.contentAttachments.updateAttachmentProperties({
  id: '12345',
  attachmentId: 'att67890',
  body: {
    id: 'att67890',
    type: 'attachment',
    title: 'report-final.pdf',
    version: { number: (attachment.version?.number ?? 1) + 1 },
  },
});

List — v2

const attachments = await v2.attachment.getPageAttachments({ id: 12345 });

for (const attachment of attachments.results ?? []) {
  console.log(attachment.title, attachment.fileSize, attachment.mediaType);
}

There are siblings for every container: getBlogpostAttachments, getCustomContentAttachments, getLabelAttachments, and getAttachments for the whole site.

Download

const bytes = await v1.contentAttachments.downloadAttatchment({
  id: '12345',
  attachmentId: 'att67890',
});

await writeFile('report.pdf', Buffer.from(bytes as Uint8Array));

The method name is Atlassian's typo. It is kept because it is the operation id — renaming it in the client would hide which endpoint you are calling.

TIP

The download is a 302 to a signed URL; fetch follows it, so you get the bytes and not the redirect. If you would rather hand the URL to a browser, v2.attachment.getAttachmentById({ id }) gives you downloadLink.

Delete — v2

await v2.attachment.deleteAttachment({ id: 'att67890' });