Why reach for this instead of curl?

It encodes the System One request shape, quick per-question flags, and a decisions-only summary mode in one binary, so shell pipelines and agent tool definitions stay small instead of hand-building JSON.

How do I use it inside a script?

Echo the state-plus-questions JSON into the binary over stdin or pass a file flag, and add the summary flag when you want just each question's top decision with its probability.

Where do the key and endpoint come from?

The key travels in an environment variable with optional overrides for model and base URL. Documented defaults point at the standard API host and the latest Jev model.

Is it safe to embed in agent tooling?

That is its stated purpose, with a dedicated agent skill definition and use-case docs for devops, data, finance, and agent flows. Every call still needs a valid key and network access.

Clone the repository and run cargo install --path ., or cargo build --release and use the binary at target/release/jeb. It needs a TypeSafe key in the environment and is MIT-licensed.

Can I skip the JSON and use flags?

Yes. You can pass --state with the text and then --choice, --option and --noul flags, for example a next_step choice with deploy and os_setup options plus a need_backup noul. The stdin, argument and file forms remain for richer requests.

What does the decision-summary mode return?

With -d it returns just each question's top decision without the full distribution, along with the model and usage. In the documented example a noul question returns decision yes with probability 0.94, model jev-1.13.0 and input and output token counts.

CommunityJev Agents & MCP

Jeb

Command-line client for the System One API aimed at scripts, shell pipelines, and agent tool calls: pipe in a state with typed questions and get typed decisions back without a full LLM workflow.

See highlights

Open github.com

Stars: 1
Language: Rust
License: MIT
Published by: Community
Author: wnoronha
Last verified: Oct 2026

wnoronha/jeb

User repository on GitHub

View on GitHub

CLI tool to query TypeSafe AI's System One API (Jev model). Designed for scripts, shell pipelines, and agent tool calls where you need fast decisions without invoking a full LLM workflow.

Forks: 0
Last pushed: Updated Oct 2026

Repo stats from the GitHub API, cached Oct 2026.

About this project

CLI tool to query TypeSafe AI's System One API (Jev model). Designed for scripts, shell pipelines, and agent tool calls where you need fast decisions without invoking a full LLM workflow. Set your API key via environment variable: To get just the top decision per question without the full distribution: API Reference — Payload schema and response format. CLI Reference — Available flags and options. Use Cases — Examples for devops, data, finance, and agents. Skill Definition — Agent tool specification.

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

Tagscommunity github

Works forJev for Tool calling

Used inJev for Software & DevTools

Languages: Rust

Highlights

Accepts JSON over stdin, as an argument, from a file, or via quick flags with no JSON required.
Supports choice, noul, and score questions, with a summary mode returning only the top decision each.
Defaults to the standard API base URL and latest Jev model, overridable through environment variables.
Ships API, CLI, use-case, and agent-skill docs, with test, clippy, and fmt checks for contributors.

Watch out

Needs a TypeSafe key in the environment; it is a thin query client, not a hosted decision service. 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

jev-mcp

Rust stdio MCP server letting Codex offload bounded judgments to Jev: four tools for yes/no, choice, score, and batch, while Codex keeps investigation, edits, and tests.

Jev Agents & MCP#community#github

Community

ieee0824/jev-mcpRust

keel

330

A local-first macOS coding workspace that runs Claude Code, Codex, Cursor, Grok, Hermes, and pi over ACP, with a local Laya or hosted Jev selector choosing routes that the host then validates.

Jev Agents & MCP#community#jevbest

Community

codejunkie99/keelRust

decapod

236

Rust governance kernel that sits between your coding agent and the repository, turning intent into bounded specifications, recorded state, and proof-backed publication across many separate invocations.

Jev Agents & MCP#community#jevbest

Community

DecapodLabs/decapodRust

Albatross

235

Rust terminal coding harness with one TUI across local and cloud providers, per-turn cost on the status line, routed plans scored by explicit policy, real undo, and forkable session paths.

Jev Agents & MCP#community#jevbest

Community

morganlinton/AlbatrossRust

jev-cli

Rust terminal tool that asks Jev noul, choice and score questions about piped text or files and returns calibrated probabilities with stable exit codes for shell scripts, CI jobs and agents.

Jev Agents & MCP#community#jevbest#catalog-import

Community

shaharia-lab/jev-cliRust

nitro

Grok Build wrapper that asks Jev one noul per tool before the first model call of a turn, then holds the shortened tool list so the prompt cache survives.

Jev Agents & MCP#community#awesomejev#catalog-import

Community

daniel-farina/nitroRust

Back to all resources

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai

Jeb

About this project

Highlights

Questions about this entry

Why reach for this instead of curl?

How do I use it inside a script?

Where do the key and endpoint come from?

Is it safe to embed in agent tooling?

How do I install it?

Can I skip the JSON and use flags?

What does the decision-summary mode return?

More Jev resources like this

From the Jev community