Get Tags (URL)

The other half of dumbdemon's Booru-Prompter pack, DDGetTagsByURL is the version of "Get Tags" for people who'd rather copy a whole post link than dig out the ID. Paste the URL of a Danbooru or E621 post, and the node fetches it, splits the tags into the same six categories as its sibling, and hands you prompt-ready strings for whatever Danbooru-trained model you're running (Illustrious, NoobAI, Pony - tags are the native language of all of them).

Honestly, if you usually work from a browser tab you might prefer this node over the ID version just because there's nothing to transcribe. The mechanism is otherwise identical, so this article stays short - the full rundown on outputs, install and settings lives on the DDGetTagsByCode page.

Inputs that matter

• booru_url - the full post URL, e.g. https://danbooru.donmai.us/posts/1234567 or an e621 equivalent. Must be a /posts/<id> link; a tag-search or artist page will error out.
• remove_underscores, add_space, exclude_tags - same three toggles as the ID node. Underscores stay by default, comma-space joining is on (leave it), and exclude_tags strips a comma-separated list from every output.

Outputs

The same six STRING outputs as DDGetTagsByCode: all (full prompt, artist tags prefixed like by_sakimichan), tags_only (general category), plus artist, character, copyright, and meta (where the quality tags like masterpiece and highres live). Wire all into your CLIP Text Encode and queue.

One difference worth knowing

The ID version pulls from whatever booru you've configured in settings. This one fetches from whatever domain is in the URL, so you can paste an e621 link even when your default site is Danbooru and it still works - the pack treats Danbooru and E621 as the two supported sources. The catch is minor: the disk cache keys entries by post ID and tags them with your configured site, so a URL lookup of a post you already cached under the other setting will fetch again rather than reuse. You'll never notice unless you flip-flop between sites a lot.

Install and the first-run gotcha

Same pack, same steps: ComfyUI Manager → search Booru Prompter, or

cd ComfyUI/custom_nodes
git clone https://github.com/dumbdemon/Booru-Prompter

then restart. No model files, no heavy dependencies - pure Python with diskcache (and PyYAML, which ComfyUI already ships). If the nodes don't show up, pip install diskcache in your ComfyUI environment fixes it.

And the thing that trips up every new user: before any fetch works you have to set your booru User ID in ComfyUI Settings → Booru Prompter. The default placeholder rejects every request, and you'll know because all six outputs fill with the same error text instead of tags. Set your real user ID once and the pack stops nagging.