How does decision-router pick between models like haiku, sonnet and opus?

It asks Jev a Choice over the candidates in both orders and averages the answers, then applies a complexity Score as a floor; low confidence falls back to the configured model.

Can it export training data for a local routing model?

Yes. decision-router export writes labeled decisions as JSONL in the Exu dataset format, and --teacher additionally includes Jev's distribution as a soft target.

Where is the decision-router API key stored?

decision-router auth set keeps the TypeSafe key in the macOS Keychain on a Mac, or in ~/.config/decision-router/credentials.json with mode 0600 elsewhere, and TYPESAFE_API_KEY in the environment always wins.

What happens when Jev is slow or unavailable?

The router returns the configured fallback model and says why; there is a timeoutMs budget and routing never blocks your prompt, so a slow or down Jev or a missing key degrades to the fallback.

Does decision-router consider quota or context limits?

Yes. Before asking anything it removes candidates whose context window is too small or whose quota is nearly gone, reading a usage-monitor status endpoint and dropping candidates below minRemainingPercent.

How do I evaluate the router on my own prompts?

decision-router eval takes a JSONL case file of prompt and expected candidate pairs, and the report splits mistakes into too cheap, too expensive and sideways so you can see the direction of each error.

Can I use decision-router from code instead of the CLI?

Yes. The package exports decide, JevBackend and DEFAULT_CONFIG, so you pass a profile's candidates with a backend, fallback, confidence and timeout and read the chosen model from the returned decision.

CommunityJev Tools & CLIs

Decision Router

A CLI, Claude Code plugin and Pi extension that pick a model per task from one Jev call: a Choice over candidates asked in both orders plus a complexity Score that sets a minimum tier, with quota-aware filtering and fallback.

See highlights

Open github.com

Stars: 0
Language: TypeScript
License: MIT
Published by: Community
Last verified: Oct 2026

Ruivalim/decision-router

User repository on GitHub

View on GitHub

Pick the right LLM for each prompt with a calibrated decision model (TypeSafe Jev). CLI, Claude Code plugin and Pi extension.

Forks: 1
Last pushed: Updated Oct 2026

Repo stats from the GitHub API, cached Oct 2026.

About this project

Pick the right model for each prompt. A rename goes to the cheap model, a deadlock investigation goes to the strong one, and you stop paying top-tier prices for one-line edits. decision-router asks a calibrated decision model (TypeSafe's Jev) which of your candidate models fits a task, in about 300 ms. It ships as: a CLI (decision-router pick "..."), usable from any script or agent; a Claude Code plugin: /decision-router:route picks haiku, sonnet or opus and delegates the task to a subagent on that model;

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

Tagscommunity routing claude-code cli

Works forJev for Routing Jev for Classification Jev for Scoring

Used inJev for Software & DevTools

Languages: TypeScript
Platforms: CLI
Capabilities: Open sourceRequires API key

Highlights

On 30 agent-labeled prompts, adding the complexity floor raised accuracy to 93% (28/30) from 73% (22/30) for Choice only.
A Jev routing call takes about 300 ms; defaults set timeoutMs 2000 and minConfidence 0.3 with a fallback model.
Quota filter reads a usage-monitor status endpoint and drops candidates below 10% remaining.
Decisions are logged to decisions.jsonl; feedback and export produce Exu-format training data, with a --teacher soft-target mode.

Watch out

The 30-prompt comparison is described as a smoke test, not an independent benchmark. MIT licensed; requires Node 22 or newer and a TypeSafe API key.

Questions about this entry

Short answers drawn from the linked project or page.

Reactions & coverage

Posts, threads, and videos about this entry from around the web.

Facebook: Intake Router keeps diagnosis out of the answer space at the front desk.

FacebookClinicnotes

YouTube: Jev explained for routing and judging, with a comparison against LLMs.

Hacker News: Vercel on using Jev for defined decisions and GPT-6 Astra when the job also has to write or use tools.

flashbrew4 pointsTypeSafe AI Jev vs. GPT-6 AstraRead the thread on Hacker News

Hacker News: ASTGL replay chose local Laya routing over hosted Jev on one Mac Studio setup.

Jmeg8r15 pointsDoes an open-weight decision model beat a hosted one? Jev vs. LayaRead the thread on Hacker News

Related terms

Glossary definitions related to this entry.

Intent routing

More Jev resources like this

jev-opus

Claude Code plugin and CLI that runs Claude Code on Opus 5.5 and has Jev re-decide the reasoning effort when a prompt arrives and after every tool batch, sending each change as a per-message effort statement inside one unbroken prompt cache.

Jev Tools & CLIs#community#claude-code#routing

Community

WXK-AI/jev-opusTypeScript

Pi TypeSafe Router

A Pi coding-agent extension that uses Jev to classify each request and route it to one of three configured models. Supports TypeSafe, Cloudflare AI Gateway, Vercel AI Gateway and OpenRouter backends, with doctor checks, shadow mode and an off-by-default setup.

Jev Tools & CLIs#community#coding-agents#routing

Community

jekozyra/pi-typesafe-routerTypeScript

Paca

1.9k

A Go-based team workspace with optional Jev AI for agent auto-routing, task auto-fill, auto-assign, and automation condition nodes.

Jev Tools & CLIs#community#go#agents

Community

Paca-AI/pacaGo

surf

634

A CLI browser automation tool with optional Jev semantic mode: Surf observes, Jev selects the next action, Surf validates and acts, Jev checks the goal.

Jev Tools & CLIs#community#browser#cli

Community

nicobailon/surf-cliTypeScript

jeff

Go CLI for Jev with noul, choice, score, eval, rank, and models commands: rank sends one Score question per item per weighted YAML dimension in a single request and sums weight times score in code, while threshold flags turn a failed gate into exit code 10 for scripts and CI.

Jev Tools & CLIs#community#cli#go

Community

saembit/jeff-cliGo

Jev Agent Skill

Claude Code and ZCode skill that offloads classify, route, batch-screen, score and compliance-check judgments to Jev through OpenCode Zen's free tier, bundling a zero-dependency jev.py caller and a Taobao comment-triage pipeline that keeps raw items out of the agent context.

Jev Tools & CLIs#community#claude-code#skills

Community

yuyang2230/jev-agent-skillPython

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