Set Scheduler

Sampler and scheduler travel as a pair in every ComfyUI workflow, and WorkflowX splits the difference: Set Sampler handles one, Set Scheduler handles the other. If you've ever changed karras to normal in three KSamplers and missed one, this node exists to make that a single edit. Publish a scheduler under a key, and every Get Scheduler in the graph resolves the same choice.

Yes - no output, same as the rest of the Set family. Set Scheduler registers the chosen scheduler under key in WorkflowX's config state, and Get Scheduler reads it back. The dropdown is validated against your live ComfyUI install (the schema shows the standard nine: simple, sgm_uniform, karras, exponential, ddim_uniform, beta, normal, linear_quadratic, kl_optimal), so you can't typo a scheduler into existence.

The config angle

The reason to use this instead of just typing karras into each sampler is the same one that drives every node in the Get Set Go family: values are config-scoped. WorkflowX's Config SelectorX assigns native groups to named configurations, so a "Fast" config can publish simple while a "Quality" config publishes karras under the same key. Flip the config and the whole graph's scheduler changes with it - the workflow-profile pattern the pack advertises, applied to the most second-guessed dropdown in sampling.

Inputs:

• key (STRING) - the name to match in Get Scheduler. Empty keys are rejected.
• value - the scheduler dropdown, live-validated.

Install

The whole pack is one install:

cd ComfyUI/custom_nodes
git clone https://github.com/haroonaslam/WorkflowX-Configurator

or search WorkflowX Configurator in ComfyUI Manager, restart ComfyUI, hard-refresh. No dependencies, no models.

Wire a Get Scheduler with the same key into a KSampler's converted scheduler input. If you're setting up per-config values, put the Set inside a group you've assigned in Config SelectorX.

The pairing gotchas

• Get Scheduler wants Set Scheduler. The two keyed channels (sampler and scheduler) are separate. A Set Sampler won't satisfy a Get Scheduler and vice versa - pair them by type and by key.
• Keys are case-sensitive. karras as a key is not Karras.
• A valid scheduler still needs a compatible sampler. The node guarantees the scheduler name exists, not that it's sensible with your sampler or checkpoint. On modern flow-matching models the model's own suggested scheduler is usually the right one - if your results look off after "optimizing" the scheduler, the default was probably correct.

It's the quiet half of a two-node pattern. If you've already got Set Sampler in a workflow, adding Set Scheduler keeps the pair honest.