ComfyUI-Jev
ComfyUI custom nodes that use Jev to select candidates, judge conditions, score text, extract numbers and choose local Skills.
Overview
ComfyUI-Jev adds custom nodes to ComfyUI so that image-generation graphs can branch on Jev judgments. The Jev Interpret node selects, judges or scores text and extracts numbers; Jev Skill Choice picks local SKILL.md files for a request; an OpenRouter Text node generates text or selection candidates that can then be judged.
It targets ComfyUI users who want natural-language conditions inside a workflow without writing code or parsing LLM prose. The decision model sits inside the graph: text or JSON state goes in with an instruction, and a typed result (a choice, a probability, a score or a number) comes out and feeds downstream nodes. TypeSafe is the default provider, and the README describes switching the Jev nodes to OpenRouter as an alternative. The project is marked as work in progress.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which candidate best matches the instruction?(inferred) | — |
| Yes / no | Does the text satisfy this condition?(inferred) | true · false |
| Score | Where does this text land on the given scale?(inferred) | — |
| Extract | Which number does the text state? | — |
How it works
- Feed text or JSON (optionally images) into a Jev Interpret or Skill Choice node
- Write a natural-language instruction and options
- Node calls TypeSafe (or OpenRouter) and returns a typed result
- Route or parameterise downstream ComfyUI nodes with the result
Strengths
- No-code typed branching inside ComfyUI graphs
- Installable from ComfyUI Manager
- Provider can be switched between TypeSafe and OpenRouter
Caveats
- Marked work in progress
- API keys typed into nodes are saved into workflows and history
- Requires a hosted API key
Getting started
Install ComfyUI-Jev from ComfyUI Manager (or git clone into custom_nodes), set TYPESAFE_API_KEY and restart ComfyUI.At a glance
- Type
- Service
- Status
- prototype
- Language
- Python
- Runtime
- ComfyUI custom nodes
- Has code
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
Custom nodes for using Jev's text interpretation and judgments in ComfyUI. Use natural-language instructions to select candidates, evaluate conditions, score text, or extract numbers, then pass the results to other nodes. Jev judgments use the TypeSafe API by default.
logicrw/awesome-jev-projectsJev text interpretation and judgments for ComfyUI.
hellogumbo/awesome-jev
Frequently asked questions
What can the nodes decide?
They select a candidate, judge a condition, score text, extract numbers, or choose which local Skills apply to a request.
Does it need a TypeSafe key?
By default yes; the Jev nodes can alternatively be pointed at OpenRouter with an OpenRouter key.
How is this different from an LLM text node?
The Jev nodes return typed values that downstream nodes can route on, rather than free text you have to parse.
Related projects
CLI that has Jev judge a video's subtitles segment by segment, then cuts a highlight reel and a filler-free full version.
ProjectbetaPythonChrome extension that converts a page's WebMCP tool schemas into Jev questions to pick a tool and fill its arguments from what you type.
Browser extensionprototypeJavaScriptBlender extension where Jev ranks about 2,500 operators against a plain-language request via a two-stage Choice search.
ProjectprototypePythonProof-of-concept MCP tool dispatcher that routes a natural-language command and extracts arguments using only Jev Choice and Noul.
MCP serverprototypePythonComposio agent SDK with an optional TypeSafe provider that uses Jev to shortlist and pick tools and closed-set arguments.
ProjectproductionTypeScriptPython multi-agent runtime with native Jev support: agents ask typed questions via a decision client separate from the text model.
LibraryproductionPython