Browser display

Ask an LLM for a report, a comparison table, or anything formatted, and you'll usually get markdown back - headers, bullet lists, bold text, maybe a table. Dump that into ComfyUI's plain text preview box and you get a wall of # symbols and pipe characters instead of anything readable. Browser display fixes that specific annoyance: it takes markdown or raw HTML and actually renders it, so the LLM's formatting effort isn't wasted.

How it works

Wire in a string of markdown or HTML - typically the direct output of one of the party's LLM nodes - and this node renders it as a real page rather than treating it as plain text. It's an output node, meant to sit at the end of a branch where you want a human to actually read the result, not to feed something further downstream.

The inputs and outputs that matter

• md_or_html - the content to render. Feed it markdown for reports and summaries, or raw HTML if you (or the LLM) generated something more structured.
• is_enable - toggle rendering off without removing the node from your graph.
• One output: log - a string output alongside the render, useful if you want to also log or pass along what was displayed.

How to install it

• ComfyUI Manager - search "comfyui_LLM_party", install, restart.
• Manual - cd ComfyUI/custom_nodes && git clone https://github.com/heshengtao/comfyui_LLM_party, then restart.

Then pip install -r requirements.txt from inside the comfyui_LLM_party folder, in your ComfyUI Python environment (python_embeded\python.exe -m pip install -r requirements.txt on portable Windows installs). No extra account or key needed for this one - it's a display node, not something that calls out to a service.

Common issues & troubleshooting

Nothing renders, or it looks like raw text. Check what's actually feeding md_or_html - if the upstream LLM node's output isn't cleanly markdown (say, it's wrapped in extra quoting or a JSON envelope from a tool-calling response), you'll see the literal characters rather than a rendered page. Strip that wrapping upstream if needed.

HTML renders but looks broken or unstyled. This node renders what you give it; it doesn't add its own stylesheet or sanitize markup. If your HTML depends on external CSS or scripts, don't expect those to load - keep the HTML you feed it self-contained if you want consistent output.

Only part of a long response shows up. If the LLM's response got truncated upstream (hit a token limit, or a tool-calling response only returned partial content), that's a generation-side issue, not a rendering one - check the LLM node's max-tokens setting before assuming this node dropped content.

The pack won't load at all. Separate from this node, comfyui_LLM_party has a real, current install issue where a Manager install completes but nodes still fail to import - usually a CUDA/PyTorch mismatch with no clear compatibility list published. If Browser_display isn't showing up in your node browser, check that first.