Enviral CLIP Name

Load CLIP's clip_name field is a dropdown, and ComfyUI doesn't let you wire a string into a dropdown. Fine when your workflow always uses the same text encoder - annoying when you're building a graph that should switch encoders based on something decided earlier, or iterating over a list of encoders.

Enviral CLIP Name is the bridge: a dropdown of your CLIP/text-encoder files that also accepts a linked string and emits the exact native combo value Load CLIP wants. You can keep picking from the menu, or you can let another node drive it - same output either way.

How it works

Its option list comes from folder_paths.get_filename_list("text_encoders"), so it shows every text encoder ComfyUI has registered, including anything you've added through extra_model_paths.yaml. When you feed it a linked string, it normalizes and matches in order: exact, then case/path-insensitive, then unique path suffix. A bare filename resolves even if the full registered path has subfolders. If nothing matches, it fails validation before the loader runs and suggests the closest valid names in the error message.

That last bit matters more than it sounds. Wrong encoder names usually surface as "this model loaded but results are garbage." Here you get a hard, readable validation error instead.

Input and output

• clip_name - dropdown or linked string.
• Output clip_name - native combo, wired into Load CLIP's clip_name.

Two sockets, nothing else to configure.

Install

Ships in the Enviral Design Node Pack. Via ComfyUI Manager, search "Enviral Design Node Pack"; or:

cd ComfyUI/custom_nodes
git clone https://github.com/EnviralDesign/comfyUI-enviral-design-node-pack

then restart. No extra dependencies.

Realistic use

This is the kind of node you install once for a workflow and forget. Its natural home is a "model flavor" switcher graph - one Custom Combo (or a string splitter) decides the architecture, and Enviral CLIP Name, Enviral Diffusion Model Name, and Enviral CLIP Type all follow the same string. If you never automate encoder selection, you'll never miss it; but it's tiny, dependency-free, and makes those branching graphs dramatically less brittle.