Skip to content
Metatext Decide — typed decisions for your agents, with a probability on every answerGet early access
Metatext
Sign Up

typesafe-mcp

Go CLI and MCP connector exposing an evaluate tool for Jev and other System One models to Claude Code, Claude Desktop, Codex, Hermes and pi.

CLIbetaMIT
View on GitHub ↗

Overview

System One Connector, installed as the evaluate command, gives AI agents direct access to System One models. An agent asks something like "is this urgent?" or "which team owns this?" and gets back a number or an option it can use in an if statement, rather than a paragraph. A setup command finds Claude Code, Claude Desktop, Codex and Hermes and registers the tool with each; pi is also supported.

It is not tied to one provider: besides TypeSafe's Jev (directly or via OpenRouter), it can run self-hosted open models such as CLM and LAYA, or hosted ones such as Liquid AI's d1. It is written in Go, MIT licensed, and installs with a shell script on macOS and Linux.

What the decision model decides

TypeQuestionAnswers
Yes / noIs this urgent?yes · no
ChoiceWhich team owns this?billing · technical · sales

How it works

  1. Install the evaluate CLI.
  2. Run evaluate setup mcp with a TypeSafe or OpenRouter key.
  3. Agents call evaluate with state and questions.
  4. They branch on the returned probabilities.

Strengths

  • Auto-registers with several agents.
  • Works with Jev and open System One models (CLM, LAYA, d1).
  • Single static binary.

Caveats

  • Hosted use needs a TypeSafe or OpenRouter key.
  • Open models must be served separately.

Getting started

curl -fsSL https://raw.githubusercontent.com/itsmostafa/system-one-connector/main/install.sh | sh && TYPESAFE_API_KEY=your-key evaluate setup mcp

At a glance

Type
CLI
Status
beta
Language
Go
Runtime
CLI + MCP server
Has code
yes
License
MIT

Use cases and integrations

How curated lists describe it

  • MCP ecosystem: Go-based CLI and MCP server exposing an evaluate tool that routes typed decisions to Jev with automatic configuration for Claude Code, Claude Desktop, Codex, and pi.

    yibie/awesome-jev

  • An MCP server that lets Claude Code, Claude Desktop, Codex and Pi ask Jev typed questions.

    logicrw/awesome-jev-projects

  • MCP connector that gives agents access to Jev decisions.

    cobanov/awesome-jev

Frequently asked questions

Which agents does setup find?

Claude Code, Claude Desktop, Codex and Hermes; pi is also supported.

Can I use open models?

Yes, such as CLM and LAYA self-hosted, or Liquid AI's d1 hosted.

Do I need TypeSafe access?

No, an OpenRouter key also works for Jev.

Related projects

  • jev-rs

    Rust engine that answers noul, choice and score from any LLM in one prefill via logprobs, wire-compatible with /v1/systemone and MCP.

    ModelbetaRust
  • jev-mcp

    MCP server exposing Jev-style decision models served locally by Ollama's /v1/systemone as classify, yes/no and score tools.

    MCP serverbetaTypeScript
  • System One Connector

    Go CLI and single-binary MCP server giving Claude Code, Codex, Hermes and pi an evaluate tool backed by Jev, d1, CLM or Laya.

    MCP serverbetaGo
  • Building with Jev★ 155

    Agent skill for writing and improving programs that call Jev: question design, state structure, thresholds and diagnosing wrong answers.

    Agent skillproduction
  • askjev

    Unofficial hosted or local Jev MCP server giving agents a quick second opinion: it picks yes/no, scale or choice and returns probabilities.

    MCP serverbetaTypeScript
  • decidealot

    Docker service exposing local Laya, Von, a CLM head or hosted Jev through one TypeSafe-compatible HTTP and MCP endpoint.

    MCP serverbetaPython