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

jev-mcp

Standalone MCP server for T3 Code with one jev_route tool where Jev picks a model and reasoning-effort combination for a task.

MCP serverbeta
View on GitHub ↗

Overview

This jev-mcp provides Jev-based model routing for T3 Code as a standalone MCP server, kept outside T3 Code's repository and database so official updates do not conflict. It has one tool, jev_route: given a task summary, a policy and the model and reasoning-effort combinations the caller may use, it asks Jev to choose one.

The answer is advisory and is used to pick a model for a subagent or a new thread; it cannot switch the current thread's model. Each request id is recorded in a local SQLite store before Jev is called, so repeats return the stored decision and reusing an id with different input fails. The repository also includes a cutover tool that migrates a database from the earlier Jev fork of T3 Code to official builds, with dry run, backup and checks, plus a bench-ledger-router agent skill.

What the decision model decides

TypeQuestionAnswers
ChoiceWhich model and reasoning-effort combination should handle this task under the given policy?—

How it works

  1. The agent summarises a task and lists allowed model and effort combinations
  2. jev_route records the request id, then asks Jev to choose
  3. The stored decision is returned (idempotent on repeated ids)
  4. The agent uses it for a subagent or new thread

Strengths

  • Idempotent decisions stored in SQLite
  • No npm dependencies
  • Separate from T3 Code, so updates do not conflict
  • Tests included

Caveats

  • Specific to T3 Code users
  • Advisory only; cannot switch the running thread's model
  • Requires Node 24+ and a TypeSafe key

Getting started

git clone https://github.com/nyattoh/jev-mcp.git and register it per docs/configure.md; requires Node 24+ and TYPESAFE_API_KEY.

At a glance

Type
MCP server
Status
beta
Language
TypeScript
Runtime
Node MCP server
Has code
yes
Has tests
yes

Use cases and integrations

How curated lists describe it

Frequently asked questions

What does jev_route decide?

Which allowed model and reasoning-effort combination should handle a summarised task.

Can it change the model of my current thread?

No. It is advisory and used for subagents or new threads.

Who is it for?

T3 Code users who previously ran the Jev fork or want model routing as an MCP tool.

Related projects

  • Helm

    Orchestrator that probes which coding agents are installed and logged in, then has Jev pick the best one for a task.

    Agent skillbetaPython
  • jev-agent-hooks

    Hooks for Claude Code, Codex and pi that use Jev to suggest a skill each turn, plus a rule gate for subagent model routing.

    HookbetaJavaScript
  • jev-conductor-router

    MCP server and CLI where Jev decides if a coding task needs a Conductor cloud workspace, how hard it is, and which harness and model run it.

    ProjectbetaTypeScript
  • loadout

    Chooses the model tier, skills and MCP servers for each coding-agent session with two Jev requests, then launches Claude Code or Codex.

    MCP serverbetaPython
  • jev-codex-router-skill

    Portable Codex skill that asks Jev to choose a model and reasoning tier for independent subtasks, with confidence gating.

    Agent skillbetaPython
  • jev-pilot

    Claude Code plugin that asks Jev before each turn for reasoning effort, subagent model, strategy and skill, with a decision log.

    Agent skillbetaTypeScript