PromptDBBridge(TJ)

PromptDBBridge (TJ) is the boring middleman that makes the TJ PromptDB system usable, and that's the nicest thing you can say about a plumbing node. Its entire job: take one TJ_PROMPT_PIPE - the bundle that PromptDBSave writes and PromptDBLoader reads - and unfold it into 11 individual outputs you can actually plug things into. You never really reach for this node on its own; you reach for it because you want a saved prompt row's values here on the canvas, not stranded at the loader across the graph.

How it works

A pipe is just a tuple of prompt settings that travels on a single socket. The Bridge unpacks it into its fields and, crucially, re-emits the same pipe on its own pipe output - so you can chain one bridge after another, or branch off mid-stream, without ever losing the original bundle.

The outputs, all from the pipe's fields:

• positive_prompt, negative_prompt (STRING) - straight into your CLIP Text Encode.
• model_name (STRING), seed (INT), steps (INT), cfg (FLOAT), sampler_name, scheduler (STRING), extra_settings (STRING), source_path (STRING).

Here's the genuinely clever bit: sampler_name, scheduler, and model_name are typed as wildcards, so you can wire them directly into the COMBO dropdown inputs of a KSampler or checkpoint loader. A plain STRING output can't do that - ComfyUI will refuse the connection. The wildcard type is the pack's workaround, and it's the reason the Bridge can feed a whole sampler config from a saved row instead of just showing you text.

There's also a get_name optional input so a wireless TJ provider can feed the pipe without a visible wire, and an auto_set toggle that republishes every unpacked output as a wireless provider under PDBB_* names. Turn that on and you can scatter prompt values around a large workflow with zero wires - which is, again, the whole design religion of this pack.

Installing it

Same pack, same drill:

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

or ComfyUI Manager → ComfyUI-TJ_NODE → restart. The PromptDB family needs openpyxl, which the pack auto-installs.

When you'd actually use it

The intended pattern: PromptDBSave logs a run to Excel → PromptDBLoader browses that log and spits out the selected row as a pipe → the Bridge sits next to whatever node actually consumes the settings and fans them out. In a section-based workflow you'd run one pipe across the canvas and drop a Bridge right before the sampler, keeping the graph clean and the values local to where they're used. If you've never touched the PromptDB system, this node will look like 11 outputs in search of a purpose - it only starts paying rent once you're logging and reusing prompt rows for real.