AIHub Expose Project Config Float

AIHub Expose Project Config Float is the continuous-value member of the pack's project-config quartet. Where the integer sibling handles counts, this one reads a floating-point number - a CFG-like strength, a denoise amount, a speed parameter - out of the current AIHub project's configuration and hands it to your workflow as a FLOAT output.

Quick refresher if you haven't met the project system yet: AIHub projects let a client bundle state around a workflow or a chain of workflows - a config plus files - and that state is managed by the client, not by ComfyUI. A workflow participates when its AIHubWorkflowController has a project_type set and the workflow runs against an already-initialized project. The config itself can be a config.json in a folder, or whatever the client prefers; the README is explicit that storage is the client's business. This node is just the read side of that contract.

What you set

• field - the config key, using dots to reach sublevels. quality.strength maps to {"quality": {"strength": 0.8}}. This is the input you'll actually edit.
• default - the value returned when the field is missing or empty. Defaults to 0.0. For a strength or denoise value, a 0 default is often a landmine - if the client hasn't set the field yet, your workflow will run with the parameter effectively off. Choose a default your workflow can survive.
• id - the protocol field id (alphanumeric, _, -).

No label, no tooltip, no sort index - like its siblings, this is a hidden read, invisible to the end user.

The pattern to know

The float node pairs naturally with AIHubActionSetProjectConfigFloat: one workflow computes a value (say, the CFG that worked last time) and writes it into the project; this node in a later workflow picks it back up. That's the whole "project state across workflows" trick in miniature. And because the client can set it without any user-facing field, it's also how the client silently passes its own preferences into your graph.

Install & gotchas

Standard pack install: ComfyUI Manager → search ComfyUI-aihub-workflow-exposer, or git clone https://github.com/otavanopisto/ComfyUI-aihub-workflow-exposer into custom_nodes, then restart. No requirements.txt, no model downloads.

Same gotchas as the other config nodes, worth repeating because they're the ones that actually bite: no project context means the default comes back, silently; and the stored value's type must actually be a float. Store an integer or a numeric string and you'll get the default instead of a coercion. If your float keeps snapping back to the default while you're sure the field is set, check the schema the writer node used - dots versus different nesting is the usual culprit.