☁️Mini Story Generator

This one's a bit different from the rest of the pack, and its category name gives it away: it lives under "mini-party" rather than the model-chain or tools categories where the serious building blocks sit. Give it a theme and an API key, and it hands back a short story plus a character description - no loader node, no separate tool wiring, no prompt engineering required. It's the closest thing this pack has to a "does it work?" test node, and honestly a decent quick sanity check to run right after installing, before you go build a five-node agent pipeline and can't tell whether a problem is your graph or your API key.

There are only two required fields: theme (multiline text, defaulting to a Chinese sample about a tortoise-and-hare race) and model_name (defaulting to gpt-4o-mini, a sensible cheap default for a throwaway test). base_url and api_key are optional - leave them blank and it'll fall back to whatever you've set centrally in the pack's config.ini, or fill them in directly on the node if you want this one workflow to use different credentials than the rest of your graph. is_enable is the standard on/off toggle you'll see across most of this pack's nodes, letting you disable it without deleting it from the graph. Two outputs come back: story, the generated text, and character, a separate description - presumably meant to feed a character-consistency step elsewhere in a larger workflow (an image generator, a TTS voice pick, whatever you're building), though the node itself doesn't do anything with character beyond handing it to you.

Should you build a real workflow around this? Probably not as your main tool. It's a fixed, single-purpose task - write a story from a theme - with none of the flexibility of the pack's actual LLM node, which lets you set your own system prompt, manage conversation history, and wire in tools. Think of Mini Story Generator less as a production node and more as the pack's equivalent of a "hello world": something bundled to prove the install works and give new users an instant result with zero setup, in the spirit of what the README calls "the simple start of the first LLM application required by ordinary students." If you like what it produces, the natural next step is rebuilding the same idea with the LLM + LLM_api_loader pair, where you actually control the system prompt instead of whatever's baked into this node.

Installing it is the same as installing the whole pack: search "comfyui_LLM_party" in ComfyUI Manager, or git clone https://github.com/heshengtao/comfyui_LLM_party into custom_nodes and restart, then run pip install -r requirements.txt from inside the pack folder using ComfyUI's own Python (on a portable install, that's python_embeded\python.exe -m pip install -r requirements.txt - using your regular system pip is the classic mistake).

If it fails, it's almost always the API credentials, not the node - an empty or error-filled story output with no other visible error usually means the key or base_url resolved to nothing valid. Check config.ini first, then the node's own optional fields, and remember base_url needs to end in /v1/ for OpenAI-format endpoints.