Data Temp Manager

Data Temp Manager is the housekeeping node for aidec's session-variable system. Data Temp Set writes variables into ComfyUI's session memory, Data Temp Get reads them back, and Data Temp Manager is how you see what's in there and clean it up. List everything, delete one variable, delete a single array element, or wipe the whole store. Think of it as the maintenance panel for the pack's little in-memory scratch space.

It's firmly in the pack's experimental tier - the README documents the queue and loader nodes and calls the rest test pieces - so this is a debugging and tidy-up helper, not something a normal batch touches. But if you are building loops with the Data Temp nodes, it earns its place fast, because the single biggest gotcha with that session store is stale state carrying over between runs. Manager is how you inspect and reset it.

How it works

It reaches into the same session-wide store the Get/Set nodes use and performs a management action against it. There's no data flowing through your pipeline here - it's operating on the store directly and reporting what it found or did. Being able to see what variables exist, and how many, is genuinely useful when a loop is misbehaving and you suspect it's reading leftovers from a previous batch.

The inputs that matter

• action - the operation: list_all (show me what's stored), delete_var (drop one variable by name), delete_index (drop a single array element), or clear_all (wipe everything).
• var_name (optional) - which variable to act on, for the delete actions.
• array_index (optional) - which element, for delete_index.

Outputs: variables_list (the names currently in the store), status, and total_vars (how many there are). It's an output node, so you can drop it into a graph on its own just to peek at or reset the store without wiring it into anything.

Installing it

cd ComfyUI/custom_nodes && git clone https://github.com/aidec/Comfyui_TextBatch_aidec, then restart ComfyUI (or ComfyUI Manager → Install via Git URL). No models, no heavy dependencies. Category: TextBatch.

Where people get burned

• clear_all is a blunt instrument. It wipes every variable in the session store, not just the one you were thinking about - including anything another branch of your graph is relying on. Use delete_var when you only want to reset one thing.
• It's the fix for stale state, so run it deliberately. The Data Temp store persists across batches until cleared. Dropping a clear_all (or a targeted delete) at the start of a run is the clean way to make sure you're not accumulating on top of last run's data.
• Experimental, thin docs. Like the rest of the Data Temp trio, this is test-tier. Use list_all liberally while you're figuring out what your loop is actually storing.

Nobody reaches for Data Temp Manager first. But the moment your Data Temp loop starts returning results from a batch you thought you finished, this is the node that shows you why - and clears it.