Prompt Show & Locker (TJ)

If you've ever chased a prompt through four chained text nodes, squinting at tiny widgets to figure out which one actually produced the string that hit your sampler - Prompt Show & Locker (TJ) is the sanity node for you. It's a tiny inspect-and-lock utility: show a STRING value large enough to actually read, and optionally freeze it so it survives reruns no matter what upstream changes.

How it works

Text comes in one of two ways: the text widget, or a connected text_input (so a Prompt Enhancer, Prompt Studio, or any string node can feed it). The node displays the value prominently with a Copy button so you can grab it for the clipboard without selecting tiny text. Then it passes the string through on text_output so you can drop it inline in a chain without breaking the flow.

The interesting part is lock_in_prompt. When it's on, the node holds onto the value you locked - across workflow re-runs and even reloads - instead of letting whatever's upstream overwrite it on the next execution. That's the "Locker" half of the name, and it's exactly what you want when you finally get a prompt right and don't want a wireless provider or an LLM node to quietly replace it during a batch.

Both halves are wireless-capable too: get_name lets it receive from a TJ provider without a wire, and set_name publishes the locked value back out under a name, so other nodes can read it without a visible connection.

Inputs and outputs

• Inputs: text (the widget), text_input (optional STRING in), get_name (optional wireless receive), set_name (publish name), lock_in_prompt (BOOLEAN).
• Output: text_output (STRING).

It's an output node, so it counts as a workflow end point for queue purposes - you don't need anything hanging off its output for it to be "done."

Installing it

It's in the pack, which installs the standard way:

cd ComfyUI/custom_nodes
git clone https://github.com/designloves2/ComfyUI-TJ_NODE

or ComfyUI Manager → ComfyUI-TJ_NODE → restart. No special dependencies - no LLM models, no llama-cpp build. Despite sitting in the LLM category (it lives alongside the prompt-generating nodes because it's part of that workflow), this node is pure text plumbing.

The honest take

This is a deliberately small tool, and that's fine - "show the string" and "don't let the string change" are two problems worth five minutes of node each. The lock_in behavior is the real value: in any workflow where an LLM or a wireless provider feeds your prompt, one Show & Locker right before the CLIP Text Encode gives you a checkpoint you can trust, and a visual confirmation that the prompt you think you're sampling with is the prompt that's actually there. If your prompts are stable, you may never need it. If you've ever gotten a surprise in a batch, you'll keep one around.