Integer Bypasser v1.0.0 (AcademiaSD)

Let me be honest about what this node is, because the name will trip you up. "Integer Bypasser" doesn't bypass integers - the integer is the input that decides how many connected nodes stay active. It's a control hub: you wire other nodes into it and it flips them between ON and BYPASS for you, either with individual toggles or automatically from a count. And yes, it has zero outputs. That's not a bug, it's the design - this is a switchboard, not part of your data flow.

How it works. This is one of those nodes where all the real work happens in the frontend. The Python side just declares the inputs and prints a log line; the JS (from the pack's js/ folder) watches the node's input sockets, adds a toggle widget per connected input, and labels each toggle with the title of the node feeding that socket. Flip a toggle and it directly sets that upstream node's mode to ALWAYS or BYPASS in the graph - no re-wiring, no deleting cables. The schema ships four LATENT inputs, but the frontend keeps adding more as you connect nodes, so it scales past four; that's why the README talks about in1 to in5 and the node happily grows to ten.

The active_count input is the automation lever, and the author's tooltip says it exactly: "Active bypass (0 = all off, 1 = Only one, etc.)". Set 0 and everything hangs bypassed. Set 3 and the first three connected nodes stay active while the rest get bypassed. Wire an integer generator or a counter into it and you get sequencing for free - run one pass with two stages active, next pass with three, without touching the canvas.

Inputs:

• in1–in4 - connect the nodes you want to control. They're LATENT-typed in the schema, so the straightforward targets are sampler-stage nodes that output latents; the frontend inherits types as it adds more sockets.
• active_count - optional INT, 0 to 64. Drives the sequential ON/BYPASS logic.

When you'd actually reach for it: modular workflows where you want to A/B a stage - is that second pass helping or hurting? - or toggle steps on and off while keeping a clean, exportable graph. It's the kind of thing that sounds like it should be core ComfyUI and isn't, which is exactly why packs keep re-implementing it.

Gotchas worth knowing. Because the logic is frontend-only, if the pack's JS fails to load - a bad update, a JS error from another node - this thing does literally nothing but print a log line. And the bypass state lives in the graph, not in the executed workflow: the mode flip happens when you connect or queue, not as part of the data flow. Treat it as a canvas convenience, not pipeline logic you can rely on in automation. Also note the input sockets that need a connection to be meaningful: an unconnected toggle has nothing to control, so you'll want everything wired before you trust the counts.

Install. ComfyUI Manager → search comfyui_AcademiaSD → install → restart, or cd ComfyUI/custom_nodes && git clone https://github.com/AcademiaSD/comfyui_AcademiaSD. No extra Python dependencies for this node. This is a single creator's pack (the Academia SD YouTube channel) with a small footprint - no big community, so the GitHub issues page is where real fixes land.