MiniMax Prompt Batch Loader

Not every MiniMax workflow goes through an LLM. Maybe you hand-write your captions and lyrics, or you keep a folder of prompt files you've refined over months, and you want to render them all in one queue without retyping a thing. MiniMax Prompt Batch Loader is the no-LLM on-ramp: point it at a directory (or flip to manual mode), and it reads each prompt file and emits song variants with reproducible source metadata and seeds.

Modes

• folder (default) - reads every file in prompt_directory matching the extensions list (default .txt,.prompt,.md), with recursive toggling subfolder scanning. Each file is one song source.
• manual - uses the fields on the node instead. manual_title, manual_caption, manual_lyrics. The caption is where musical/production instructions go; the lyrics field is for supported section tags and lyric text. Instrumental tracks should use structural tags, not prose.

Either way, song_count (1–100) controls how many variants each source emits, and seed_mode picks between random_each_song and the reproducible increment_from_base (from base_seed). For batch rendering you can reproduce later, increment mode is the sane default; random mode is for exploration.

Outputs

All lists, mapped across your batch by ComfyUI: caption, lyrics, title, source_name, seed, run_index, and source_path. No image prompt here - this is the audio-only loader. source_path is the prompt file that produced each song, which is exactly the provenance breadcrumb the metadata sidecar wants.

Where it fits

It's the plainest entry point in the pack: LLM-free batch generation straight from files. If your workflow later graduates to an LLM, the sibling nodes (template + parser) slot in upstream and this loader becomes unnecessary - but for reproducible file-driven batches, it's the one you reach for.

Installing it

Pack install via ComfyUI Manager (search "MiniMax Music Production Toolkit") or:

cd ComfyUI/custom_nodes
git clone https://github.com/jplenio/ComfyUI-MiniMax-Music-Production-Toolkit.git
cd ComfyUI-MiniMax-Music-Production-Toolkit
python -m pip install -r requirements.txt

Dependencies: scipy, soundfile, imageio-ffmpeg, mutagen, Pillow. Restart ComfyUI and hard-refresh once.

Gotchas

The directory must live on the machine running ComfyUI, and files with unsupported extensions are silently ignored - if a batch seems short, check your extensions string and that recursive matches where the files actually are. And a caution from the pack's design: garbage prompt files produce garbage songs. The LLM branch exists to enforce the artifact-avoidance rules; this node trusts your files blindly, so the quality bar is on you.