Easy Prompt (W/ Append)

Easy Prompt (W/ Append) is the fuller sibling of the pack's Easy Prompt node. Same core idea - a CLIP text encoder that loads LoRAs from <lora:name:weight> tags written directly in your prompt - plus two extra inputs that merge in additional prompt text after you're done typing. The README calls it out for "Krita-style prompts," which is the honest reason it exists: when a Krita AI plugin is tacking its own tags onto whatever you write, you want a place to add them without editing your main prompt box.

The LoRA machinery is identical to the simple version, and it's real: the node regex-scans for <lora:...> tags, resolves each against your models/loras folder, applies them with ComfyUI's own load_lora_for_models call, then strips the tags and encodes the cleaned text into conditioning. Weight syntax matches the ecosystem standard - <lora:name:0.8> for both model and clip, <lora:name:0.8,0.5> to split them.

What the append fields do

append_positive and append_negative are the whole difference. The node merges each append field into its main prompt with a comma - "myprompt", "extra tags" becomes "myprompt, extra tags" - and empty fields are skipped entirely. So you can keep your hand-written prompt clean while a plugin or another part of the workflow injects stuff after it. The merge happens before LoRA parsing, so you can even put <lora:...> tags in the append fields.

Inputs and outputs

• clip (CLIP, required) - the text encoder.
• positive, negative, append_positive, append_negative (all STRING, multiline) - four boxes, two you write, two that get tacked on.
• model (MODEL, optional) - feed it or your LoRAs only patch the clip side.

Outputs: MODEL, Positive, Negative - same shape as any CLIP text encode node, ready for a sampler.

Should you use this or the simple one?

Honest answer: if nothing in your workflow is appending prompt fragments, use the simple Easy Prompt node. The append fields are dead weight you'd stare at wondering what they're for. If you are driving a Krita AI diffusion workflow - or anything where prompt text comes in from somewhere besides your fingers - this is exactly the right tool, because "my prompt, plus whatever the plugin wants" is a pain to express with plain text boxes.

Same gotchas as the simple version: wire model in or your LoRA tags barely do anything; a typo'd LoRA name leaves the raw <lora:...> tag in your conditioning with only a console warning to tell you. And like any CLIP encode, this is for CLIP-based models (SD 1.5, SDXL, Illustrious, Pony lineage), not the newer LLM-encoder models.

Installation

Part of ComfyUI-Jax-Nodes. In ComfyUI Manager, search "ComfyUI-Jax-Nodes" and install, then restart. Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/Kebolder/ComfyUI-Jax-Nodes

Restart and it's there. No pip dependencies, no models to download, and the pack needs a reasonably current ComfyUI (it uses the newer extension API, so late 2025 or later). It's a small pack and not heavily updated, but what's here is stable.