Feature request: Add auto complete/ a easier way to copy trigger words for Lora

[static holly]

note: original issue from https://github.com/mcmonkeyprojects/SwarmUI/issues/394. idk why the issue is closed without any reply, but for notification sanity let's make a discussion thread here instead.

Now the ways of adding trigger words for lora can be concluded into three ways:

adding as prompt
copying the trigger words by:
switching display format of LoRA tab into Card modes; find your LoRA card; highlight the trigger phrase; copy; paste it on prompt
In any display format of LoRA tab, click 'Edit metadata' of your LoRA; copy values of trigger phrase; click cancel; paste it on prompt
type it by yourself
However, the methods have their own weakness, especially when the LoRA contains the concept of (a) character(s), with trigger phrase separated on clothes:

The first method always fill in everything, blocking users to make characters with different looks.
The second method is complex as:
Users seem to be forced to use Card display format if they want to copy trigger phrase without making popups; and
Copying at 'Edit metadata' dialog is dangerous as user is too dumb to keep it intact.
User is too dumb to remember trigger phrase for the third method.
This issue suggests a new way that allows users to input trigger phrase they needed, not necessarily all phrases as in . We may take examples from the AI drawing service in the market. For example:

in CivitAI, the trigger words are listed, allowing users to one click copying them, and then paste them in the prompt

in tensor.art, the tirgger words are listed allowing users to add them directly to the prompt

With this, trigger phrase is also suggested to be a list of "phrases" when stored in the metadata, allowing selecting subsets of phrases, which the subset is provided by the LoRA author.

[modest sedge]

what do you mean closed without any reply

jay answered pretty sufficiently I think

[static holly]

coz in my view jay only answers pupphelper's comment, but my point was not addressed

I do mentioned <trigger> keyword in my original post, and I clearly know that this syntax adds ALL trigger words from LoRAs with those provided

[modest sedge]

oh

well, your request I think will still be covered by one of the key feature targets listed in the readme https://github.com/mcmonkeyprojects/SwarmUI?tab=readme-ov-file#status full detail "Current Model" display in UI, separate from the model selector (probably as a tab within the batch sidebar?)

[static holly]

great! let's see if this feature handles well for users to add trigger words easier

[gloomy ether]

This is a nice feature, however many loras have mutiple options or multiple trigger words, which you would want to pick and chose rather than applying all trigger words available in the lora. No rush looking forward to the upcoming model selector as mentioned here. Thanks for all the great work team!