Skip to main content
All examples live under examples/ and run with a real API key.

​
SystemOne

Asks all three question types (Choice, Score, Noul) about a support message and prints the typed answers plus token usage.

​
ListModels

Prints every model the API currently serves, with its release date and description.

​
Common Jev patterns

These three Go examples follow patterns that show up repeatedly across the Jev community: a spam classifier, a PR labeler, and a game move picker. Each also links a few personal-favorite projects using the same pattern in other languages. Those links are unrelated to this Go SDK, added only because they’re fun to see the pattern working elsewhere.

awesome-jev

Community directory of 800+ Jev projects. Every linked project below was found there — credit to its maintainers and contributors for cataloging them.

​
Spam filter

A Noul question per message, thresholded in code. Not related to this Go SDK, just a personal-favorite Jev use case: same shape as moderation and antispam bots like Jev-Moderation-Bot or scam-shield.
Live third-party demo, embedded from scam-shield-seven-ecru.vercel.app. Not affiliated with this SDK.

​
PR labeler

A Choice question over a diff, labeling its conceptual scope (security, feature, refactor, docs) instead of counting changed lines. Not related to this Go SDK, just a personal-favorite Jev use case: same pattern as jev-pr-labeler and commit-miner.

​
Game move picker

The game engine owns state and rules; one Choice question per tick picks the move. Not related to this Go SDK, just personal-favorite Jev use cases: the pattern behind dozens of Jev-plays-a-game demos, including jev-tetris and Jev Pac-Man.
Live third-party demos. Not affiliated with this SDK.

​
Personal-favorite Jev use cases

None of the four projects below are written in Go, and none use this SDK. They’re added only because they’re fun, rounded up by Moritz Kremb in this thread. Full credit to each builder; these are their projects, unrelated to typesafe-sdk-go beyond calling the same API.

​
Keystroke oracle launcher

dabit3/launcher by Nader Dabit — a native macOS launcher that judges intent on every keystroke. Type “the pdf I just downloaded” and Jev ranks the newest PDF top, at full confidence, in about 100 ms.
dabit3/launcher: typing 'the pdf I just downloaded' ranks the newest PDF top at 100% confidence in 154 ms
Not related to this Go SDK, just a personal-favorite Jev use case.

​
Context compaction

tamaratran/fast-jev-compaction by tamara tran — a Claude Code plugin that replaces lossy LLM-summarized compaction with Jev decisions: every tool call and result is scored, stale ones are dropped, everything kept stays verbatim. Not related to this Go SDK, just a personal-favorite Jev use case.

​
Ultrafast browser agent

browser-use/jev-ultrafast by Browser Use — Jev picks the operation and element for each browser action instead of a full LLM generation pass. Zürich → London on Google Flights in 7.1 seconds, one natural-language goal. Browser Use jev-ultrafast: a real Google Flights search completing in 7.1 seconds Not related to this Go SDK, just a personal-favorite Jev use case.

​
Plain-English Postgres

realZachi/pg-jev by Zachi — ask a Postgres table questions in plain language; Jev turns the question into a query. See it live at pgjev.zachi.dev (not embeddable here, the site blocks iframes). Not related to this Go SDK, just a personal-favorite Jev use case.

View source on GitHub

Full source for every example.