☁️Mini Summary Generator

Sometimes you don't need an agent, a persona, and a memory chain - you just need a wall of text turned into three sentences. Mini Summary Generator is exactly that: feed it text, get a summary back, nothing else on the node to configure beyond the model and your credentials.

It's part of comfyui_LLM_party's "mini-party" family - self-contained, single-purpose LLM utilities that skip the pack's usual loader/persona/memory assembly. If you've ever built the full three-or-four-node LLM chain in this pack just to do one small text transform, mini-party nodes are what you actually wanted.

Inputs that matter. input_str is the text to condense - required, no default, and there's no length cap here beyond whatever your model's own context window allows, so this scales from a paragraph to a genuinely long document. model_name defaults to gpt-4o-mini, a sensible choice for summarization specifically: it's cheap, fast, and summarizing doesn't usually need frontier-model reasoning. base_url and api_key are optional the same way they are everywhere in this pack - leave them blank and the node falls back to whatever's configured in config.ini. is_enable is the standard bypass toggle to skip the call without deleting the node from your graph.

One output: output_text (STRING), the summary itself - wire it straight into whatever's next, a Save Text node, a display node, or another LLM call that needs the condensed version rather than the original wall of text.

Where it fits. Reach for this whenever a downstream step in your workflow needs the gist of something rather than the whole thing - condensing scraped article text before feeding it to a persona-driven writing node, summarizing a long RSS entry, or trimming conversation history down before it goes into a prompt with limited room. It's the kind of node you'll wire in without thinking twice once you know it exists.

How to install it. Ships with the full pack:

cd ComfyUI/custom_nodes
git clone https://github.com/heshengtao/comfyui_LLM_party.git

Then pip install -r requirements.txt from inside the project folder using ComfyUI's own Python, and restart. Or search "comfyui_LLM_party" in ComfyUI Manager and install from there directly. No local model or GPU cost specific to this node - it's a thin API wrapper.

Common issues & troubleshooting. An empty or garbled output_text almost always traces back to credentials - check api_key/base_url on the node, then check config.ini if both are left blank. If the summary is longer than you expected, that's a prompt-quality thing more than a node limitation: input_str is passed through with a fixed summarization instruction, so if you need a specific length or format, you'll get more control from this pack's full LLM node with your own system prompt than from this mini version, which trades configurability for zero setup. And separately from this node's own behavior: this pack's combined requirements.txt is large enough that a fresh install occasionally fails to import cleanly - if the node doesn't show up in your browser at all, check the ComfyUI console for an import error before assuming you've misconfigured something here.