Show And Save Anything 🐑

Every serious ComfyUI graph eventually needs a node whose only job is "let me actually see what's on this wire." rgthree-comfy's Display Any node is the community's default answer to that, and Show And Save Anything is MieNodes' version of the same idea - except it adds one thing the popular equivalents mostly don't: a plain-text log file on disk, so you don't have to keep the browser tab open to catch what ran.

It accepts absolutely anything on its anything input - a string, a number, an image tensor's shape, whatever a prompt generator or LLM connector spat out - converts it to a readable string, prints it where you can see it in the graph, and (if save_to_log is on, which it is by default) appends that string to a log file. It's an output node, which in ComfyUI terms means it's meant to sit at the end of a chain rather than feed further nodes, though it does also return a STRING output if you want to keep piping the same value onward.

Why you'd actually reach for it. The pack's own README workflow example is the clean case: load an image, generate a caption with Florence2, then run that caption through an LLM connector to produce an English and a Chinese prompt side by side - and use this kind of node to display both at once so you can read the result without digging through the console. Anywhere you've got an LLM-generated string (a translation, a rewritten prompt, a hash from StringHash) and you want to eyeball it before it goes further, this is the node you drop at that junction.

Inputs that matter. anything - required, accepts any type, this is the value you're inspecting. save_to_log - a boolean, default true, controlling whether the value also gets written to disk. log_file_name - optional, default show_anything.log, the file it appends to.

The log file is the differentiator. If you're running a workflow headless, through the API, or on a remote GPU box where you're not watching the UI in real time, the printed value in the node itself is useless to you - you weren't looking. The log file is what saves you there: point log_file_name at something descriptive if you're running multiple workflows that all use this node, since by default they'll all append to the same show_anything.log and you'll end up with everything interleaved. Splitting log files per workflow, or per stage of a pipeline, is a cheap habit that pays off the first time you need to debug a run you weren't watching live.

Install. Search "ComfyUI_MieNodes" in ComfyUI Manager, or cd ComfyUI/custom_nodes && git clone https://github.com/MieMieeeee/ComfyUI-MieNodes and restart ComfyUI. No dependencies, no models - this is a pure Python utility node.

Troubleshooting. If nothing shows up, first check anything is actually connected - an unconnected required input on an output node will just error the run rather than silently display nothing. If you expected a log entry and don't see one, check save_to_log is still true (easy to flip off while testing and forget to flip back) and check the log file's actual path - it's written relative to wherever ComfyUI's working directory is, which trips people up on Docker or remote setups where "next to the README" and "where the process actually runs" aren't the same directory. And if you're seeing a wall of unrelated entries in the log, that's multiple workflows sharing the default filename - give each one its own log_file_name.