To KSampler Settings

Every KSampler in ComfyUI asks for the same five things: a seed, a sampler, a scheduler, steps, and a CFG. To KSampler Settings is a single node that bundles all five into one socket, so you can define your sampling settings once and hand the whole bundle to anything that wants them. It's the settings half of the "one wire carries everything" pattern, and it's built to feed straight into the pack's To DuckComfy Globals node.

The mechanism is unapologetically simple: the node collects the five values into a 5-tuple - (seed, sampler_name, scheduler, steps, cfg) - and sends it out as a BASIC_PIPE. Its companion node From DuckComfy Globals unpacks that tuple on the other end. Nothing is executed or validated here; this is a carrier, not a sampler. The actual sampling still happens in your KSampler.

The inputs that matter

• seed (INT, default 0) - with the usual control_after_generate fixed/increment/randomize widget attached, so you can still randomize per run from here.
• sampler_name - all 44 samplers ComfyUI knows, from euler to dpm_2_ancestral and beyond. This list tracks the engine, so if a new sampler lands in core ComfyUI it shows up here too.
• scheduler - the nine vanilla schedulers: simple, sgm_uniform, karras, exponential, ddim_uniform, beta, normal, linear_quadratic, kl_optimal. Note this is deliberately the plain vanilla list, not the extended set that Efficiency Nodes adds - the pack keeps those extras in a separate "Cast Scheduler To Detailer" node for compatibility. If you want AYS SDXL or a GITS scheduler, it's not in this dropdown.
• steps (INT, default 60) and cfg (FLOAT, default 6.5).

That default of 60 steps at CFG 6.5 is SDXL-era thinking. If you're feeding this into a Flux workflow, you'll want CFG around 3.5 and far fewer steps - this node is not model-aware, it just carries whatever you dial in. Set your values for the checkpoint you're actually running.

The output

• ksampler_settings (BASIC_PIPE) - the 5-tuple, ready for To DuckComfy Globals or anything else that accepts a BASIC_PIPE.

How to install it

It's part of duckcomfy personal nodes, a personal utility pack that author duckcomfy re-implements to keep dependencies out of his workflows. Install via ComfyUI Manager (search "duckcomfy personal nodes"), or:

cd ComfyUI/custom_nodes
git clone https://github.com/duckcomfy/duckcomfy_personal_nodes

Restart after cloning. No models, no dependencies - the pack's pyproject.toml declares an empty dependency list, which is rare for a custom node and nice for your sanity.

Gotchas

The BASIC_PIPE type name is borrowed from Efficiency Nodes, and the pack's scheduler lists are made compatible with that ecosystem - but don't assume the tuples are interchangeable. DuckComfy's settings bundle is its own 5-tuple shape, so treat this as a closed loop with To DuckComfy Globals unless you've verified the other end expects the same layout. And remember: changing the seed here randomizes every sampler fed from this bundle, which is exactly what you want for reproducibility, and exactly what will bite you if you forget the connection and wonder why every branch renders identically.