Get Prompt From Batch
The 'just give me prompt #3' node
- prompt
Batch Text is the node that runs all your prompts; Get Prompt From Batch is the one that runs one of them, on purpose. It's the small companion node you reach for when firing off the whole list is wrong - you want to re-queue prompt #3 that came out great, A/B two prompts against each other, or step through a list one queue press at a time. Same pack, same zero-dependency setup, opposite job.
How it works
While Batch Text pushes a list out, this node pulls one item back out. It marks its input as list-aware (INPUT_IS_LIST), so it receives the entire prompt list in a single call, then grabs the entry at index. Two behaviors matter here:
- It's 0-based. Index 0 is the first prompt. This trips up people more than it should.
- It wraps around: the actual index used is
index % len(prompts_list). Ask for index 99 on a three-prompt list and you quietly get prompt 0. That's a deliberate safety net - no out-of-range crash mid-workflow - but it can look like a bug if you're not expecting it.
Inputs and outputs
Only two inputs, and only one you'll actually touch:
prompts_list- wire this to theprompts_listoutput of Batch Text (the "Batch Text (Prompt Loop)" node in the same pack).index- an INT, default 0. Set it as a widget to pick a prompt, or drive it from another node if you're building something clever.
The single output, prompt, is a plain STRING - connect it anywhere you'd connect a prompt, which is usually CLIP Text Encode.
Why you'd actually use it
The honest use case is manual control. Add it to a batch workflow, expose the index widget, and you get a "pick which prompt to run" dial without duplicating text everywhere. It also pairs naturally with the pack's image-side counterpart, Get Image From Batch, using the same index value on both - now you're looking at exactly one prompt×image combination, which is great for debugging a matrix workflow before you let the full 3×5 loose. Drive that index from a simple INT widget and you've got a poor-man's iteration rig: bump the index, queue, bump, queue.
Installing it
It's bundled with the rest of ComfyUI-batching-nodes (MIT, by Hahihula). ComfyUI Manager: search "ComfyUI Batching Nodes" and install. Or manually:
cd ComfyUI/custom_nodes
git clone https://github.com/Hahihula/ComfyUI-batching-nodes
Restart ComfyUI after either path. There's no pip dependency to install and no model file to fetch - the pack only uses what ComfyUI already ships, so this node appears the moment the pack loads.
Gotchas worth knowing
- The wrap-around behavior means an out-of-range index never errors - which is nice for robustness, infuriating if you expected "prompt 99" and got prompt 0.
- An empty or missing list returns an empty string rather than failing, so check the
countfrom Batch Text if you're seeing blank prompts. - It's a picker, not a scheduler. If your goal is "run each prompt automatically in sequence," you don't need this node at all - just wire the list straight through and let ComfyUI fan it out. Use this node when you want fewer than all of them.
Inputs (2)
| Name | Type | Default | Description |
|---|---|---|---|
| prompts_list | STRING | — | |
| index | INT | 00–10000 | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| prompt | STRING | — |