User repository on GitHub
Probabilistic decisions for Python. Use Jev or bring your own provider; crawl with Playwright.
- Forks
- 0
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
CommunityJev Repos & SDKs
Decido
A Python layer over a small decision-provider contract that adds validated distributions, batching, budgets, ranking and exportable records, with Jev as the first provider.
- Stars
- 1
- Language
- Python
- License
- MIT
- Published by
- Community
- Author
- yairshy
- Last verified
- Oct 2026
About this project
Ask typed questions about evidence, keep every probability, and use the results in ordinary Python. Start with Jev or supply your own decision provider. Decido adds validated distributions, batching, request budgets, ranking, and exportable decision records around a small provider contract. It also includes optional web crawling and local MCP tools. Get a key from the TypeSafe console. Jev requests use your account and may incur charges. Environment setup.
Quoted from the project's README. Read the full README.
Works forJev for Classification
Used inJev for Software & DevTools
- Languages
- Python
Highlights
- A session batches your questions into one call and returns typed answers through binary, choice and ordinal accessors, each with its distribution.
- It ships a TypeSafe provider but is not tied to it: any object implementing the provider protocol can be passed to the session or model directly.
- The plain install includes a lexical provider, so the bundled ranking example runs with no network access and no TypeSafe key.
- Runnable examples cover support routing to review, document ranking with independent relevance probabilities, and separating reported changes from missing data.
Watch out
This is an alpha release under MIT and the README warns the API may change before 1.0. It is also explicit that model probabilities are not automatically calibrated and that independent questions do not form a joint distribution, so do not treat a set of answers as a coherent probability model. The keyless provider is a lexical token-overlap heuristic, useful for exercising the API but not a semantic judgment. Anything routed to Jev bills your TypeSafe account. Most operational detail lives in separate docs pages rather than the README, including the JSON request format, provider contract, crawling and MCP setup. 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.
Independent. The library defines a small provider contract and ships one implementation backed by Jev, plus a lexical one. Implement the protocol yourself and pass your object straight to the session or model, which is the intended extension path. Switching providers is configuration, not a rewrite of your decision code.
No, and the README says so in the same breath as the API tour: model probabilities are not automatically calibrated, and independent questions answered in one call do not form a joint distribution. If you need coherence across questions, you have to model it yourself; the library preserves each distribution but does not compose them.
Yes. Installing without the provider extra and running the rank command executes a bundled token-overlap example through the lexical provider, which makes no network calls. The README is clear that its scores are heuristics rather than semantic judgments, so it proves the plumbing works, not that the decisions are good.
Make missing evidence a separate outcome instead of a middling one, which the docs push throughout and the evidence-factors example demonstrates by distinguishing reported changes, unchanged conditions and missing information. The support-routing example takes the same idea further, sending uncertain cases to human review rather than forcing a class.
Validated distributions, batching of questions within provider limits, request budgets, ranking, and exportable decision records, wrapped around a small provider contract. It also ships optional web crawling and local MCP tools. The provider does the deciding; Decido is the bookkeeping and typing around it.
Three runnable scripts: support routing that classifies a request and sends uncertain cases to review, document ranking that selects useful documents with independent relevance probabilities, and evidence factors that distinguish reported changes, unchanged conditions and missing information. Each is meant to be edited rather than run as-is.
It is an alpha release under MIT and the README warns the API may change before 1.0. The project is also explicit that its tests verify contracts and mechanics only, and that it claims no model-quality or financial-performance advantage. So treat it as plumbing to build on, not a validated decision system.
More Jev resources like this
34GitHub stars
Drop-in Jev client wrapper that averages Choice predictions across option orderings in one request for stabler probabilities.
Jev Repos & SDKs#community#jevbest#catalog-import
Community
TypeLLM/pijevPython
TypeLLM/pijevPython45GitHub stars
Go SDK plus a pipeline-friendly CLI for typed decisions, covering single questions, batched question maps, retries, and line-oriented filtering over stdin.
Jev Repos & SDKs#community#jevbest#catalog-import
Community
mattn/go-jevGo
mattn/go-jevGo16GitHub stars
Swift library and CLI for asking Jev typed Choice, Noul, and Score questions from Apple platforms, with answers returned alongside calibrated probabilities in one parallel request.
Jev Repos & SDKs#community#jevbest#catalog-import
Community
d-date/swift-jevSwift
d-date/swift-jevSwift9GitHub stars
Go SDK for the TypeSafe API that asks typed Noul, Choice, and Score questions about a state and returns probability distributions with confidence and typed accessors.
Jev Repos & SDKs#community#jevbest#catalog-import
Community
Tangerg/typesafe-sdk-goGo
Tangerg/typesafe-sdk-goGo8GitHub stars
Zod 4 companion that adds Jev meaning checks to shape checks: one parseAsync call sends every semantic rule as noul questions in a single request and turns probabilities into ordinary Zod issues.
Jev Repos & SDKs#community#jevbest#catalog-import
Community
jomatsu/zod-jevTypeScript
jomatsu/zod-jevTypeScript8GitHub stars
Early-alpha Haskell DSL for building Jev question packets as expressions with inferred types, exact request rendering, and exhaustive typed answer handlers.
Jev Repos & SDKs#community#jevbest#catalog-import
Community
inanna-malick/jev-dslHaskell
inanna-malick/jev-dslHaskellFrom the Jev guides
Original write-ups on what this entry is for.
Closed weights have not stopped the ecosystem from rebuilding the shape in the open. API-compatible servers, retrained heads on small models, and inference-time tricks that skip generation entirely.
Jev Repos & SDKs#open-models#research#local
Read article
From the Jev community
Posts from builders shipping with Jev right now.