User repository on GitHub
Probabilistic control flow for Ruby and Rails - powered by TypeSafe's Jev
- Forks
- 0
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
CommunityJev Integrations & Adapters
Hunch — carldaws
Ruby control-flow helper that asks plain-English questions about text or app data via a System One model and returns probabilities, picks, or ordered ratings for use directly in code.
- Stars
- 19
- Language
- Ruby
- License
- MIT
- Published by
- Community
- Author
- carldaws
- Last verified
- Oct 2026
About this project
Hunch lets you ask questions about text or application data and use the answers in your Ruby code. You can get a probability, choose between options, or rate something on a scale. Questions are written in plain English; answers come back as Ruby values. Hunch asks a System One model, such as TypeSafe's Jev, which answers with calibrated probabilities rather than text. It is also available for TypeScript as @carldaws/hunch. Point Hunch at a System One endpoint. Jev is available through OpenRouter:
Quoted from the project's README. Read the full README.
Works forJev for Classification
Used inJev for Software & DevTools
- Languages
- Ruby
Highlights
- chance returns 0 to 1, with possibly, likely, probably, and almost_certainly thresholds plus custom levels.
- pick chooses among symbol or described options; rate returns a level with numeric position for ordered scales.
- Hunch.decide asks several questions about the same data in one API call with per-question answers.
- SystemOne backend speaks noul, choice, and score; Stub backend supplies answers and records calls for tests.
Watch out
Point Hunch at a System One endpoint; a missing key raises when you ask. Rescue Hunch::APIError for timeouts, rate limits, server errors, and bad answers. Auth, validation, and configuration errors are not APIErrors. Copy reviewed against the repository README on 2026-10-04, not from running it.
Questions about this entry
Short answers drawn from the linked project or page.
Parsing gives you text; Hunch gives Ruby values with calibrated probabilities: floats from chance, symbols from pick, levels with positions from rate. Threshold helpers, one-call batching, and a stub backend make branching and tests direct.
The Rails examples rescue Hunch::APIError around validations so an API failure adds no validation error. Other validations such as email presence still apply, and routing or triage callers can fall through the same way.
Assign a Stub backend with per-question answers and an optional fallback; it records calls for assertions. Supply probabilities or booleans for chance, symbols or hashes for pick, and symbols or positions for rate.
Use pick for unordered categories such as billing, support, and sales. Use rate for ordered levels such as calm, frustrated, and livid, where position and comparisons against levels matter.
Add gem "hunch" and configure a backend in an initializer: Hunch::Backends::SystemOne.new with url "https://openrouter.ai/api/alpha/decisions", api_key from ENV["OPENROUTER_API_KEY"] and model "typesafe/jev-1.13". Timeout is 5 seconds to read the response, open_timeout 2 seconds to connect, and max_retries 2 for 429, 5xx and timeouts with backoff that honours Retry-After. A missing key raises the moment you ask. There is also a TypeScript port, @carldaws/hunch.
Fixed values you can also redefine by level name. Hunch.possibly? is a chance of 0.25 or more, likely? is 0.5, probably? is 0.75 and almost_certainly? is 0.93. To add your own, set it in configuration: Hunch.configure { |c| c.levels[:definitely] = 0.99 } gives you Hunch.definitely?. The same block reconfigures the built-ins, for example c.levels[:probably] = 0.8.
One result object with an accessor per question id, and several accessors for the same one. From the documented example: result.urgent is the raw probability 0.92, result.urgent? is the boolean at or above the :probably threshold, result.team is the chosen symbol :technical, result.team_probabilities is the full hash {billing: 0.08, technical: 0.85, sales: 0.07}, and result.mood.level is the chosen rating symbol.
Other listings of this project
Forks and copies of this project in the directory. This entry is the canonical one.
1GitHub stars
Browser agent that takes one Jev decision per action on top of agent-browser and only calls an LLM when confidence drops, escalating instead of guessing, with values, origins and irreversible clicks all guarded in code.
Jev Demos & Experiments#community#awesomejev#catalog-import
Community
HAR5HA-7663/hunchPython
HAR5HA-7663/hunchPython4GitHub stars
A CLI and agent skill that asks plain-English questions across a repository, sends each in-scope chunk to the provider, returns scored source locations to investigate, and runs recurring review rules on diffs.
Jev Agents & MCP#community#jevbest#catalog-import
Community
Kelbie/hunchTypeScript
Kelbie/hunchTypeScript4GitHub stars
Python library that maps Jev's Choice, Score and Noul onto verbs like classify, score, check, where, extract, verify and rank, keeping data shape in and out, batching one request per distinct row and caching answers.
Jev Integrations & Adapters#community#jevbest#catalog-import
Community
steven-shoemaker/hunchPython
steven-shoemaker/hunchPythonMore Jev resources like this
19GitHub stars
Rails decision layer that evaluates noul, choice and score questions about app state in one Jev request, with per-question confidence floors that gate auto-action versus human review.
Jev Integrations & Adapters#community#jevbest#catalog-import
Community
Ray-Hughes/jevalynRuby
Ray-Hughes/jevalynRuby19GitHub stars
Ruby gem that models Jev as field predicates: feels and decide bind to model attributes with shared definitions, thresholds, confidence gates, batching, and test tapes.
Jev Integrations & Adapters#community#jevbest#catalog-import
Community
Qew7/jev-feelsRuby
Qew7/jev-feelsRuby69GitHub stars
Provider-neutral TypeScript runtime for System One decisions: write choice, noul, and score code once, then run it on TypeSafe Jev, local Reflex, any compatible /v1/systemone endpoint, or a custom provider.
Jev Integrations & Adapters#community#jevbest#catalog-import
Community
iamaamir/system-oneTypeScript
iamaamir/system-oneTypeScript38GitHub stars
File-first project notebook where many agents share context in .context/ files while Jev only judges the next step and records it as JSON.
Jev Integrations & Adapters#community#jevbest#catalog-import
Community
pedrocivita/tocketTypeScript
pedrocivita/tocketTypeScript41GitHub stars
Typed Python planner runtime that turns batched Jev action distributions into executable probability trees and graphs for finite decision tasks.
Jev Integrations & Adapters#community#jevbest#catalog-import
Community
Chuf-H/jev-treePython
Chuf-H/jev-treePython34GitHub stars
Self-hosted agent runtime that adds an optional Jev judgment layer to memory filtering, lesson dismissal, request classification, and skill selection, with inspectable context and execution.
Jev Integrations & Adapters#community#jevbest#catalog-import
Community
matrixorigin/AstraRust
matrixorigin/AstraRustFrom the Jev community
Posts from builders shipping with Jev right now.