MD: Universal Context Bus

If you've built an ACE-Step audio workflow with MD Nodes, you know the pain this node exists to solve: SceneGeniusAutocreator hands you ten separate outputs - genre, lyrics, duration, steps, glide power, master strength, a legacy flag, an ADG angle, vocals, and a seed - and suddenly your canvas is a spiderweb of wires going everywhere. The Universal Context Bus catches them all and lets you route them as one bundle.

How it works

Every input on this node is optional and forceInput, meaning they're noodle ports you wire things into rather than widgets you type in. The bundle function collects whatever you actually connected, drops the empty ones, and packages the rest into a single BUS output of type CONTEXT (in practice, a plain dict of your parameters).

It also re-emits a few typed outputs so you can pick values off individually: GENRE (STRING), STEPS (INT), ADG_ANGLE (FLOAT), and LEGACY (BOOLEAN). If a value wasn't connected, it falls back to a sane default - steps 180, adg_angle 0.13, legacy mode True - so downstream nodes always get a valid number instead of a crash.

What it's for

Two honest uses. First, canvas hygiene: wire SceneGenius' outputs into it, then run the typed outputs to the parts of your graph that need them. Second, snapshots: because the BUS is a dict of the whole run's parameters, you can stash it, log it, or feed it to a custom node you write that wants the full context in one object. The pack's own sampler nodes accept YAML config strings, so a context dict is a natural thing to convert and push.

The straight talk

This is a tidy-up node, not a processing node. It performs no computation - no value is transformed, nothing is validated, a string goes in and the same string comes out. If you're wondering whether it's worth installing the pack for, it isn't; it earns its keep only inside an existing MD workflow where parameter bundles are flying around. But in that context, it's exactly the kind of small thing that keeps a graph readable.

Installing it

Ships with MD Nodes:

cd path/to/ComfyUI/custom_nodes
git clone https://github.com/MDMAchine/ComfyUI_MD_Nodes.git
cd ComfyUI_MD_Nodes
pip install -r requirements.txt

Or search MD_Nodes in ComfyUI Manager and restart. Note the pack installs a heavy dependency set even though this particular node only does dicts and defaults.