What exactly gets sent to Jev when I type in the palette?

The query, command titles, tab titles, working directories and session names go to whichever provider you selected. Screen text is not sent unless you enable screen_context, which is off by default, and even then only a short excerpt per tab leaves, with bearer tokens, API keys, private keys and password= or token= values masked first.

Do I need a TypeSafe account to use this at all?

No. Without a key nothing leaves your machine and local matching still answers every keystroke. With a key, the provider setting cycles Auto, TypeSafe and OpenCode Zen, and Auto uses TypeSafe when it finds a key and otherwise falls back to Zen.

How much slower is the Jev path than plain search?

Local matching is instant per keystroke; a settled query gets its Jev ranking about a quarter of a second later. The README quotes a 290 ms median for TypeSafe direct against 654 ms for OpenCode Zen on a 15-query palette test set, and ships a probe script to re-run those queries.

What is the build situation on macOS and Linux?

macOS on Apple silicon ships as an ad-hoc signed dmg with no notarization, so you right-click and Open on first launch; Xcode 26 is optional and only used to compile a Liquid Glass icon. Linux and WSLg have no installer and run from source, and Windows uses a per-user MSI whose daemon runs inside WSL2.

How do I configure the Jev provider and key?

Open the palette, run AI Settings, paste a key and press Enter. The dialog saves one key per provider in secrets.json beside the window state, runs a connection test and reports the provider, model and latency. The provider row cycles Auto, TypeSafe, OpenCode Zen. The same settings live in config.toml under [ai].

What are the two Jev providers and their latency?

TypeSafe direct uses model jev-1.13.0 with a key starting apikey_ and a 290 ms median on the palette test set. OpenCode Zen uses model jev-1.13, picks up your OpenCode login automatically and measured 654 ms. Auto uses TypeSafe when a key is found and otherwise falls back to Zen.

Does the superterminal server make network calls?

No. The AI ranking lives in the client under packages/app/src/ai/, and the server makes no network calls at all. The Rust daemon owns terminals and streams screen rows; only the client contacts the chosen provider.

CommunityJev Tools & CLIs

Superterminal

GPU-rendered terminal multiplexer whose command palette accepts plain-language queries: Jev scores the visible command, tab and session rows so the intended one rises to the top, while local substring matching still answers every keystroke.

See highlights

Open github.com

Stars: 105
Language: Rust
License: MIT
Published by: Community
Author: sonnylazuardi
Last verified: Oct 2026

sonnylazuardi/superterminal

User repository on GitHub

View on GitHub

a native multiplexer terminal for Windows, Linux and Mac

Forks: 6
Last pushed: Updated Oct 2026

Repo stats from the GitHub API, cached Oct 2026.Homepage

About this project

A GPU‑rendered, native multiplexer terminal for Windows, Linux and Mac. Rust server (superterminald) owns the terminals; a Bun 1.4.0 + React client renders them through gpuix (React bindings for Zed's GPUI) with a native Rust element. Status: implemented and in daily use. Terminals live in a background server, so closing the window never kills a program and reopening it is instant. Windows (native client against a WSL server) and macOS (Apple silicon) ship as installers; Linux and WSLg run from source.

Quoted from the project's README. Read the full README.

Tagscommunity jevbest

Works forJev for Classification Jev for Routing

Used inJev for Software & DevTools

Platforms: CLI

Highlights

The palette sends one typed question listing every row, Jev returns a probability per row, and a confident pick rises marked with a star.
Local matching runs on every keystroke and Enter always acts on the row under the cursor, so a slow answer never blocks you.
Provider auto-detection prefers a TypeSafe key and falls back to OpenCode Zen, with a connection test reporting model and latency.
Terminals live in a background Rust server, so quitting the window leaves programs running and restores tabs and scroll positions.

Watch out

Building from source needs rustup with a pinned toolchain, Bun, and a patched gpuix submodule; the README recommends a release server build because a debug build costs roughly 40 MB per busy surface at default scrollback. The macOS build is ad-hoc signed and not notarized, so Gatekeeper fires on a downloaded copy, and Linux has no installer. Jev is optional and without a key nothing leaves the machine, but the latency figures quoted are the author's own medians over a 15-query probe. 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.

More Jev resources like this

pyGPT

A desktop AI assistant with a Jev / System One inline plugin for classification, routing, verification, and scoring without chat-style output.

Jev Tools & CLIs#community#python#desktop

Community

szczyglis-dev/py-gptPython

horosa-skill

452

An AGPL MCP server and JSON-first CLI exposing 110 astrology and metaphysics techniques from Horosa to local AI clients, running offline on macOS and Windows with SQLite-backed call history.

Jev Tools & CLIs#community#jevbest

Community

Horace-Maxwell/horosa-skillPython

332_lab-jev-chat

168

Android and Windows assistant that reads the chat already on your screen through accessibility nodes or OCR, has Jev judge intent, danger level and urgency in one call, ranks three drafted replies, and fills the input box.

Jev Tools & CLIs#community#jevbest#kotlin

Community

Liyucheng1997/332_lab-jev-chatKotlin

changelog.earth

165

An unofficial changelog for Earth that renders recent reporting as game patch notes. A scheduled GitHub Actions job collects stories, Groq picks headlines for worldwide relevance and writes the patch titles, and the result is committed as saved editions the site, API and RSS read.

Jev Tools & CLIs#community#jevbest

Community

byalex33/changelog.earthJavaScript

JevIntent

Chat-host plugin that turns a long press on a message into judgments for intent, emotion, urgency and reply posture, with a per-conversation relationship that changes how the answer reads.

Jev Tools & CLIs#community#jevbest

Community

Nisaka520/JevIntentJava

doc-router

Rust CLI that checks each PDF page for a usable text layer, extracts text pages locally, sends only the rest to an OCR model via LiteLLM, and merges a page-ordered result.

Jev Tools & CLIs#community#jevbest#catalog-import

Community

misbahsy/doc-routerRust

From the Jev guides

Original write-ups on what this entry is for.

Jev on the Gateways: Vercel, Workers AI, and OpenRouter

Oct 12, 2026

The request shape barely changes across surfaces; the surrounding contract does. When to call the API directly and when a gateway, a Worker binding, or a router entry earns its extra hop.

Jev Tools & CLIs#gateway#vercel#cloudflare

Read article

Exporting Decision Models: GGUF, ONNX, and Device Runtimes

Oct 10, 2026

Taking a decision checkpoint out of Python: what the GGUF and browser WebGPU exports preserve, what they change, and the parity evidence the model cards publish.

Jev Tools & CLIs#open-models#local#onnx

Read article

Records at Rest: Documents, Inboxes, and Risk Queues

Sep 28, 2026

Jev labels a record that already exists. Code moves, flags, or holds it. How the document, mail, and transfer tools keep the action out of the model.

Jev Tools & CLIs#classification#email#security

Read article

Back to all resources

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai