Get Dict Conditioning

Get Dict Conditioning fetches a CONDITIONING - an encoded prompt - back out of a DICT by name. It's the retrieval partner to Set Dict Conditioning, and it's how a prompt you stashed early in a workflow reaches a sampler that lives far away on the canvas. Instead of running a conditioning wire clear across a big graph, you stash it in the shared dict bag under a key and pull it out right where the sampler needs it.

This is where the dict approach earns its keep for multi-prompt setups. Base prompt, style prompt, a region-specific prompt - stash each under its own key, carry the one DICT along, and getDict whichever conditioning a given branch calls for. It's a naming-based alternative to the fixed positive/negative slots of a BASIC_PIPE, and it shines when the number of prompts is more than two and varies by branch.

How it works

Like the other typed Get Dict nodes, it does two things beyond a plain lookup (both from the README): the default is optional, and the retrieved value is type-checked. If the key is present and holds real conditioning, you get it. If the key is missing, or what's stored isn't conditioning, you get the default instead - no crash.

The inputs and outputs that matter

• key (required, text) - the name the conditioning was stored under. Match the Set Dict key exactly.
• DICT (required) - the bag to look in.
• default (optional) - a CONDITIONING to fall back on if the key isn't found or isn't the right type.

One output: CONDITIONING. Wire it into a sampler's positive/negative input, or onto a pipe via Op. Edit Basic Pipe.

Installing it

No weights, no heavy deps - it's a lightweight pure-Python pack. ComfyUI Manager → Install Custom Nodes → search "antrobots ComfyUI Nodepack" → Install → restart, or clone it:

cd ComfyUI/custom_nodes
git clone https://github.com/antrobot1234/antrobots-comfyUI-nodepack

then restart. Find it under antrobots-ComfyUI-nodepack/dicts/get.

Common issues

For object types like conditioning, the optional default matters more than it does for numbers. If the key is missing and you didn't supply a default, the node's automatic fallback is essentially "nothing" - and a sampler handed empty conditioning will error or produce garbage. So either make sure the key genuinely exists in the dict, or wire in a real fallback conditioning. Don't leave the default empty and hope.

The usual dict traps apply: keys are strings, so a typo silently returns the default rather than raising an error, and the DICT input is required. When a piped-through prompt "disappears," check the key spelling on both the Set and Get ends first. It's a small one-person pack, and the README's support model is to open a GitHub issue for real bugs - the author welcomes them.