User repository on GitHub
A Go client for the TypeSafe AI's System One API and its model, Jev.
- Forks
- 3
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.Homepage
CommunityJev Repos & SDKs
Jev — kataras
Go client for the TypeSafe System One API and the Jev model. Send one state with named Noul, Choice and Score questions and read back a typed answer, a probability for every option and a confidence. Ships a token-bucket rate limiter, a JavaScript-SDK-style retry policy and a full set of error sentinels.
- Stars
- 5
- Language
- Go
- License
- MIT
- Published by
- Community
- Author
- kataras
- Last verified
- Oct 2026
This is one of several Jev listings for the same project. The canonical entry is Jev — mayank953.
About this project
A Go client for the TypeSafe System One API and its model, Jev. Jev does not write text. You send a state (a string, or JSON such as a ticket, a log, or the current state of a program) and one or more named questions. The call returns a typed answer and a probability for every question. Adding questions to the same call does not add a round trip. This module is not affiliated with TypeSafe AI. The official clients are the JavaScript SDK and the Python SDK. The HTTP contract is the System One API.
Quoted from the project's README. Read the full README.
Works forJev for Classification
Used inJev for Software & DevTools
- Languages
- Go
Highlights
- One dependency, golang.org/x/time/rate, paces every attempt including retries against the per-account request and token limits.
- The default model is the jev-latest alias, which moves; WithModel pins a version and ListModels shows what the account can call.
- Retries 408, 429 and 500 through 599, first wait 500ms doubling to 5s with up to 25 percent subtracted at random, preferring Retry-After-Ms.
- Sentinels for config, request, response, timeout, connection, 400, 401, 403, 404, 422, 429, 529 and 5xx, plus an *APIError with status, request ID and message.
Watch out
The README states plainly that this module is not affiliated with TypeSafe AI and that the official clients are the JavaScript and Python SDKs, so it is a third-party implementation of the same HTTP contract and can lag the API. Go 1.27 is required because SystemOneAs is a generic method, a high floor for one package. Pacing is an approximation: input tokens are estimated from body size before the call and settled from usage.input_tokens afterwards, and the README notes TypeSafe adjusts the limits without notice. Debug logging writes headers and bodies as sent, and those bodies contain your state; only Authorization, Cookie and X-Api-Key are redacted. 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.
Because SystemOneAs is a generic method that decodes the same JSON straight into a struct of your type, and that is what sets the floor. JSON names in that struct are case-sensitive. If you want to avoid the toolchain upgrade, the non-generic path is SystemOne plus resp.Noul, resp.Choice and resp.Score, which needs nothing beyond the standard library and golang.org/x/time/rate. The README notes TypeSafe adjusts the documented limits without notice and that enterprise plans get higher ones, so pin your own figures with WithRateLimit.
No. The README says the module is not affiliated with TypeSafe AI and names the JavaScript and Python SDKs as the official clients, pointing at the System One HTTP contract as what this package implements. The retry policy here is therefore described as matching the official JavaScript SDK rather than defined by the provider. The wire format also makes deliberate choices other Go clients do not: JSON is encoded with encoding/json/v2 so angle brackets and ampersands are sent as themselves rather than escaped, because the text the model reads would otherwise not be the text you passed.
Effectively yes, and the README is upfront about it. POST is retried on the grounds that a System One call only evaluates, but every attempt still spends input tokens. There is also no budget across attempts unless you set RetryPolicy.MaxElapsed; the context you pass always applies, and cancellation during a wait stops the call without a retry. The ordering the client documents is that the rate limiter is the first line of defence and retries are the second, which is why it paces before every attempt rather than only after a 429.
Yes. TYPESAFE_LOG_LEVEL can be debug, info, warn, error or off; info logs one line per attempt, while debug also logs headers and bodies, and the README notes bodies are logged as sent and contain your state. Authorization, Cookie and X-Api-Key are replaced with [redacted], but request content is not, so treat debug as something you switch on deliberately rather than leave on in an environment that handles real tickets.
Go 1.27 or newer with go get github.com/kataras/jev; the only dependency is golang.org/x/time/rate.
Each account is limited to 1,200 requests per minute and 250,000 input tokens per second. The client waits on a token bucket before every attempt including retries, and clients on the same key should share one limiter.
It reads TYPESAFE_API_KEY by default and defaults to jev-latest, and explicit options win over environment variables. Set WithModel("jev-1.13.0") when a later run must hit the same model.
More Jev resources like this
45GitHub 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-jevGo9GitHub 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
Unofficial Go client for the TypeSafe System One API that sends a state plus typed Noul, Choice, and Score questions in one call and returns typed answers with model, usage, and request ID.
Jev Repos & SDKs#community#awesomejev#catalog-import
Community
devbackend/jevgoGo
devbackend/jevgoGo4GitHub stars
Community Go client for the TypeSafe System One evaluation API with raw structs and a builder that produce the same EvaluationRequest, defaulting to jev-latest.
Jev Repos & SDKs#community#jevbest#catalog-import
Community
kisshan13/typesafe-ai-goGo
kisshan13/typesafe-ai-goGo1GitHub stars
Unofficial, independently maintained Go client for the TypeSafe AI System One API with no third-party dependencies: choice, score and noul questions, typed accessors that error instead of returning zero, and per-call timeout, retry and header overrides.
Jev Repos & SDKs#community#jevbest#catalog-import
Community
cole-gillespie/typesafe-goGo
cole-gillespie/typesafe-goGo2GitHub stars
Community Go client for the TypeSafe AI System One API covering the same surface as the official Python and TypeScript SDKs, using functional options, context cancellation, sentinel errors, `log/slog`, no dependencies outside the standard library, and optional Langfuse tracing.
Jev Repos & SDKs#community#jevbest#catalog-import
Community
fgn/jevgoGo
fgn/jevgoGoFrom 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.