Which tool should I use for routing versus guardrails?

Use jev_gate for allow, confirm, or block screening before a risky action, jev_classify for routing, jev_score for ordered risk or quality, jev_check for yes-no gates, and jev_decide for several typed questions at once.

How does key setup and endpoint selection work?

Set TYPESAFE_API_KEY or its aliases. A hosted jv_live key routes to the hosted gateway automatically, other keys use the official endpoint, and JEV_BASE_URL can point at a custom gateway.

Do I need to clone or build the server locally?

It runs from GitHub with npx and no clone or build, then registers through a Claude Code command or an MCP client config with command, args, and environment key.

What prevents malformed or hallucinated tool output?

Each tool is a thin wrapper over one Jev call with the answer type fixed by the request, so Jev cannot emit an invalid type. Your agent still decides where to call it and how to enforce the result.

What does each tool return?

jev_classify returns a choice with confidence and per-option probabilities; jev_check returns a probability and a likely flag; jev_score returns a position on your ordered scale; jev_gate risk-screens an action into allow, confirm or block with a risk score; jev_decide runs many typed questions in one round trip.

How do I install it for Claude Code?

Run claude mcp add jev -e TYPESAFE_API_KEY=your_key -- npx -y github:codaaiteam/jev-mcp, or add the same command, args and env block to any MCP client config. It runs straight from GitHub with npx, so there is no clone or build.

How is the endpoint chosen, and can I route it through a gateway?

A hosted jv_live key routes automatically to the hosted gateway, anything else uses api.typesafe.ai, and JEV_BASE_URL overrides both to route through your own gateway such as Vercel AI Gateway, OpenRouter or Cloudflare. JEV_MODEL defaults to jev-latest and can be pinned.

CommunityJev Agents & MCP

Jev MCP — codaaiteam

MCP server that exposes Jev classify, score, yes-no check, risk gate, and multi-question decide tools to coding agents through one Jev call per tool.

See highlights

Open github.com

Stars: 3
Language: JavaScript
License: MIT
Published by: Community
Author: codaaiteam
Last verified: Oct 2026

codaaiteam/jev-mcp

User repository on GitHub

View on GitHub

MCP server for Jev (TypeSafe AI's System One model) — give any agent typed, calibrated decisions: classify, score, check, gate risky tool calls. Try free: jevtypesafeai.com

Forks: 1
Last pushed: Updated Sep 2026

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

This is one of several Jev listings for the same project. The canonical entry is Jev MCP — jkudish.

About this project

An MCP server that gives any agent (Claude Code, Codex, Cursor, Pi, …) fast, typed, calibrated decisions from Jev, TypeSafe AI's System One model — classify, score, check yes/no, and gate risky tool calls, all as one API call under the hood.

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

Tagscommunity github

Works forJev for Scoring Jev for Tool calling Jev for Classification

Used inJev for Software & DevTools

Languages: JavaScript

Highlights

Offers jev_classify, jev_score, jev_check, jev_gate with allow, confirm, and block, plus jev_decide for many questions in one round trip.
Detects hosted versus official keys automatically and supports explicit gateways such as Vercel, OpenRouter, or Cloudflare.
Installs via npx with Claude Code and generic MCP client configs and pins models through JEV_MODEL.
Documents guardrail, model-tier routing, and publish-gate examples with choice, confidence, probability, and risk fields.

Watch out

Needs a valid Jev key; a hosted gateway key only works at that gateway and not at the official endpoint. 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

quicksilver

118

A Claude Code skill that moves bulk judgment off Claude and onto Jev. You filter files, classify items, rank candidates or locate lines, and it returns typed verdicts, a shortlist with borderline markers, and a receipt of the tokens Claude did not read.

Jev Agents & MCP#community#jevbest

Community

UditAkhourii/quicksilverJavaScript

jev-sift

MCP server that screens files, public webpages, or inline text against a query and returns compact relevance probabilities, so the main agent opens only promising items.

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

Community

kbhuw/jev-siftJavaScript

eutrya

Standalone terminal agent that puts Jev inside its planning loop, proposing text-model steps and scoring candidates against independent rubrics before a deterministic policy executes one.

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

Community

hellozenstrategist-lab/eutryaJavaScript

jcr

Capability resolver that uses Jev to search a nested command catalog and return selected instruction context to an agent through one MCP tool, keeping the search outside agent context.

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

Community

NiazMorshed2007/jcrJavaScript

jev-mcp

MCP server exposing one evaluate tool for Jev Choice, Noul, and Score questions, plus a separate eval harness that compares Jev against LLMs on shared labeled sets.

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

Community