Skip to content
Metatext Decide — typed decisions for your agents, with a probability on every answerGet early access
Metatext
Sign Up

TypeSafe Jev + ComfyUI

ComfyUI custom node that asks Jev noul, choice or score questions inside image-generation workflows, e.g. to pick a workflow template.

Projectprototype
View on GitHub ↗

Overview

This is a custom node for ComfyUI that calls TypeSafe's Jev from inside a workflow graph. You set the state, the question type (noul, choice or score), instructions and criteria as JSON, and the node outputs the decision with probabilities and confidence. Choice supports up to 255 options and score 2 to 10 levels.

According to the list summary, the example use is routing a prompt to one of three ComfyUI workflow templates with a Jev decision instead of a keyword match. Installation is a single Python file in custom_nodes, with the TypeSafe SDK installed automatically or via pip, and an API key in the node or environment. Criteria are entered as JSON: true and false descriptions for noul, named options for choice, and an ordered list of levels for score. Answers include a calibrated confidence alongside the decision.

What the decision model decides

TypeQuestionAnswers
ChoiceWhich workflow template fits this prompt?—
Yes / noUser-defined yes/no questionyes · no
ScoreUser-defined rubric score—

How it works

  1. Add the node to a ComfyUI workflow.
  2. Provide state, question type, instructions and criteria.
  3. The node calls Jev via the TypeSafe SDK.
  4. Downstream nodes branch on the decision.

Strengths

  • Brings typed decisions into visual ComfyUI workflows.
  • Single-file install.

Caveats

  • Requires a TypeSafe key.
  • Small integration with limited documentation of examples.

Getting started

Copy typesafe_jev_node.py into <ComfyUI>/custom_nodes/, pip install typesafe-sdk, and restart ComfyUI.

At a glance

Type
Project
Status
prototype
Language
Python
Runtime
ComfyUI custom node
Has code
yes

Use cases and integrations

How curated lists describe it

  • Image generation: routes a prompt to one of three ComfyUI workflow templates with a Jev decision rather than a keyword match.

    yibie/awesome-jev

Frequently asked questions

What does the node do?

Asks Jev a typed question and outputs the decision for downstream ComfyUI nodes.

How many options can a Choice have?

Up to 255.

How do I install it?

Copy one Python file into custom_nodes and install typesafe-sdk.

Related projects

  • LangChain

    LangChain's Python framework, which includes an optional TypeSafeClassifier runnable plus model-routing and risky-tool middleware using Jev.

    PluginproductionPython
  • LangChain.js

    LangChain.js, which includes an optional TypeSafeClassifier runnable and middleware for bounded routing and tool-call checks with Jev.

    PluginproductionTypeScript
  • effect-agent

    Yielded Agent: durable Effect agent threads in TypeScript, listed for a TypeSafe decision provider for typed question sets.

    ProjectbetaTypeScript
  • eve

    Vercel's filesystem-first agent framework, whose experimental evaluate and autoModel paths default to Jev via AI Gateway.

    ServicebetaTypeScript
  • ElBruno.AI.Jev

    Community .NET 10 SDK giving one API for choice, score and assessment over Microsoft-Decision-1, Ollama or native Jev.

    PluginbetaC#
  • Building a Harness with Jev

    LangChain blog guide on wiring Jev into an agent harness as a dangerous-action classifier and model router.

    Articlediscussion