Preview as Markdown

Any node that outputs a string - an LLM response, a JSON dump, a metadata field, a prompt you built with string-concat nodes - eventually needs eyeballing. A plain text preview shows it as one long unformatted blob. Preview as Markdown renders it properly: headers, bold, lists, code blocks, whatever markdown structure the string actually contains, right there on the node.

What it's for

It's a debug and inspection tool, nothing more, nothing less. Drop it at the end of any chain that produces text and it becomes a readable window into what your workflow is actually generating. That's especially useful with anything producing structured output - a node emitting a JSON string, a metadata field like the one Spritesheet Builder outputs, or an LLM node's response - where a wall of unformatted text is genuinely hard to parse by eye but the same content rendered as markdown (headers, code fences, bullet lists) is immediately legible.

The inputs and outputs that matter

Just one field, and it's the whole node:

• source (multiline STRING, default empty) - connect any string output here, or type/paste text directly for a quick scratch note in your graph.

There are no outputs. is_output_node is true, meaning this is a terminal node - it renders for you to look at and the graph doesn't continue past it. Don't expect to chain anything off the far side.

How to install it

Search "Link Comfy Nodes" in ComfyUI Manager and install from there. Or by hand:

cd ComfyUI/custom_nodes
git clone https://github.com/Mister-Link/link-comfy-nodes
pip install -r link-comfy-nodes/requirements.txt

Restart ComfyUI. This is about as lightweight as custom nodes get - no models, no heavy dependencies specific to this one.

Common issues & troubleshooting

Expecting it to do something with the text. It doesn't validate, transform, or forward the string anywhere - it's display-only. If you need the text to continue on to another node, tap the same upstream output a second time and route it separately; don't expect Preview as Markdown to pass it through.

Non-markdown text looks plain, and that's fine. If the incoming string isn't actually formatted as markdown, it just renders as plain text with no special styling - nothing breaks, there's just nothing to render specially. That's the expected fallback, not a bug.

Multiline is a UI convenience, not a requirement. The multiline: true flag just gives the manual-entry widget a bigger text box; it doesn't restrict what kind of string can be piped in from an upstream node. A single-line string works exactly the same.

Where it actually earns its keep. This node is genuinely most useful paired with something that outputs structured or lengthy text elsewhere in a workflow - reach for it whenever you find yourself squinting at a truncated one-line string preview and wishing you could actually read the whole thing.