Any Conditioning Batch Switch

Any Conditioning Batch Switch combines several CONDITIONING inputs into one conditioning output - the conditioning socket - and it's the batch-switch sibling of the pack's single-select switches. One conditioning in, and it passes through unchanged; several in, and they get concatenated into a single stream. Where the other batch switches stack tensors, this one stacks prompt effects, and that makes it a surprisingly handy merge tool.

In ComfyUI, conditioning is a list of (embedding, metadata) entries, and a single conditioning output can hold many of them - which is exactly how regional prompting and prompt combination work. If you have two separately-encoded positive prompts, or a style conditioning and a subject conditioning, that you want the sampler to honor together, this node is a cleaner merge point than hunting for a combine node.

How it works

The node starts with conditioning_1 and grows a slot for each additional source. A single connected input passes through as-is. With several, it concatenates their conditioning entries into one list - think "union of everything you fed it," in slot order. That's the key difference from the other batch switches: conditioning has no spatial resolution to clash over, so the silent-drop behavior of the image/mask/latent batch nodes doesn't apply here. Every entry from every connected input survives.

Where you'd use it

• Prompt combination. Two separately-encoded positive prompts merged into one conditioning stream.
• Regional conditioning. Stack region conditionings into a single output for a sampler.
• Single-input passthrough. One conditioning in = plain passthrough, which is handy at subgraph edges.

Install

ComfyUI Manager → search "Tojioo Passthrough" → install, or:

cd ComfyUI/custom_nodes
git clone https://github.com/Tojioo/tojioo_passthrough.git

Restart ComfyUI. No models or pip dependencies - the pack is lean Python plus a bundled frontend, on the Comfy Registry under publisher tojioo.

Common issues

The gotcha here isn't shape - it's combination semantics. This node concatenates conditioning entries; it doesn't blend or average them. If you wire two prompts that fight each other, you get both influences, not a compromise, which is usually what you want but worth knowing. And muted or bypassed upstream sources are treated as missing and skipped (the pack-wide convention), so a branch that's muted simply contributes nothing rather than an empty entry. Keep at least one input connected or the node raises its "no inputs connected" error.