Griptape Display: Text as Markdown

When your Griptape agent answers a prompt, it usually answers in Markdown - headers, bullet lists, tables, bold text. And ComfyUI's default text display dumps all of that raw, with the ## and ** symbols intact. This node fixes that. It renders the agent's output as proper Markdown right in the node, so you read a clean formatted answer instead of a wall of syntax.

The Griptape pack made this its own node (in 2.2.12, the changelog says) precisely because the plain Display: Text node wasn't cutting it for LLM output. It's the difference between pasting an LLM's answer into a .md editor and just looking at it. For anyone building agent workflows - image description pipelines, prompt refinement, research summaries - this is the difference between "works" and "actually pleasant to read."

How it works

It's the simplest node in this pack and the most honest about it. The class subclasses the pack's Griptape Display: Text node and just changes one input type: the text field becomes a MARKDOWN-typed multiline input instead of a plain string. Feed it text, it renders it. That's the entire mechanism - no API calls, no driver, no framework.

It does have a passthrough, though: the OUTPUT returns the original string. So you can drop this node into the middle of a graph, have it render the agent's answer for you to read and forward the raw text onward to another node. Handy, since the natural instinct is to treat display nodes as dead ends - this one can be both a display and a pass-through.

Inputs and outputs

• INPUT - the text to display, multiline.
• STRING - a second, Markdown-typed input. Both feed the display; the one you wire up is up to you.
• OUTPUT - STRING, the raw text passed through.

It's marked as an output node, so it shows up as a terminal display in the graph - but as above, the passthrough means you can still wire OUTPUT forward.

Install

Standard Griptape install - ComfyUI Manager (search "Griptape") or:

cd ComfyUI/custom_nodes
git clone https://github.com/griptape-ai/ComfyUI-Griptape

The pack pulls in griptape[all] and python-dotenv via requirements.txt. You really don't need the whole framework for this node, but it ships in the pack, so you get the full dependency tree regardless. If you're on Nvidia and hit torch trouble after install, the README's fix is to reinstall torch from the cu121 index.

Common gotchas

Honestly, this node is hard to trip up. The one thing that catches people: it displays what's on the input at execution time, so if your agent output is generated mid-workflow, this node shows the result only after the workflow has run - it won't update until the upstream nodes do. And remember it's a display, not a text-transformer: if you wire OUTPUT somewhere expecting formatted text, you'll get the raw string back. There's no Markdown-to-plain-text conversion happening here.