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

feelings

Ruby gem for probabilistic conditionals like Feelings(msg).like?("urgent"), answered by Jev or other decision models.

Projectbeta
View on GitHub ↗

Overview

feelings brings the feels construct from the Probably language to Ruby. Feelings(value).like?("genuinely urgent") asks a yes/no question and returns true or false; like returns a Mood with yes, maybe and no branches, and declaring a maybe branch widens the split to 0.3/0.7 so ambiguous answers have somewhere to go. Passing at_least: sets a probability floor and returns nil for the middle zone rather than forcing a guess. most_like picks the best label from a set and returns its symbol. Descriptions and label sets can live in YAML.

The default judge is the ruby_decision_model gem, so it works with every decision model that gem reaches: TypeSafe's Jev, OpenAI's gpt-6-luna, Cloudflare's Clef, Perplexity's pplx-decider, Databricks and self-hosted System One servers. Setting OPENROUTER_API_KEY or TYPESAFE_API_KEY is enough; other vendors are chosen with an environment variable.

What the decision model decides

TypeQuestionAnswers
Yes / noIs this value like the description (e.g. spam, a security incident)?yes · maybe · no
ChoiceWhich label best fits this value (most_like)?—

How it works

  1. Wrap a value with Feelings()
  2. Call like?, like or most_like
  3. ruby_decision_model queries the configured provider
  4. Branch on true/false/nil or yes/maybe/no

Strengths

  • Idiomatic Ruby API with explicit maybe handling
  • Provider-neutral via ruby_decision_model
  • No config beyond an API key

Caveats

  • Early gem
  • Accuracy depends on the chosen provider

Getting started

gem "feelings", set TYPESAFE_API_KEY or OPENROUTER_API_KEY, then Feelings("that email").like?("genuinely urgent").

At a glance

Type
Project
Status
beta
Language
Ruby
Runtime
gem
Has code
yes

Use cases and integrations

How curated lists describe it

  • Probabilistic conditionals for Ruby, `Feelings(message).like?("genuinely urgent")`, answered by a decision model through ruby_decision_model, with descriptions and label sets in YAML.

    MrJev/awesome-jev

Frequently asked questions

Which providers work?

Jev, OpenAI gpt-6-luna, Cloudflare Clef, Perplexity pplx-decider, Databricks and System One servers.

How are uncertain answers handled?

With a maybe branch or a nil result below the at_least floor.

What is it inspired by?

The Probably toy language's feels construct.

Related projects

  • adk-go-typesafe

    Go integration bringing TypeSafe Choice, Score and Noul to Google's adk-go agents, with a typed client, routing agent and callbacks.

    ProjectbetaGo
  • Ax

    Ax, a DSPy-style multi-language LLM framework, listed with a native TypeSafe provider for Boolean, Choice, Score and raw Jev questions.

    PluginproductionTypeScript
  • Effect

    Effect TypeScript library whose @effect/ai-typesafe package maps classify, probability and rating operations to Jev questions.

    ProjectproductionTypeScript
  • instructor-php

    PHP monorepo whose Polyglot module adds a TypeSafe driver for typed Noul, Choice and Score decisions alongside LLM inference.

    ProjectproductionPHP
  • llm-typesafe

    Simon Willison's LLM plugin that adds Jev as a model for yes/no, choice and score questions from the command line.

    PluginproductionPython
  • req\_llm

    Elixir library for unified LLM calls across providers, with an evaluate interface that calls Jev via TypeSafe or OpenRouter.

    ProjectproductionElixir