alexpeclub/jev-model-router-claude-code
User repository on GitHub
- Forks
- 5
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Integrations & Adapters
Jev Model Router for Claude Code
A hook plus /jev skill for Claude Code: Jev picks the cheapest Claude tier (haiku, sonnet, opus, fable) for each prompt and estimates whether the task depends on the conversation, delegating self-contained work to a cheaper subagent.
- Stars
- 8
- Language
- Shell
- Reading time
- 2 min
- Published by
- Community
- Author
- alexpeclub
- Last verified
- Oct 2026
Works forJev for Routing
Used inJev for Software & DevTools
Highlights
- A UserPromptSubmit hook sends each prompt's text, capped at 12,000 characters, to Jev; no files or history leave the machine.
- Jev answers two typed questions: the cheapest tier and the probability the task depends on the prior conversation.
- The hook prints DELEGATE (use a subagent with the named model) or handle (stay in the session).
- Fail-open: if Jev does not answer within 5 seconds the prompt passes through unchanged.
- The /jev skill toggles the router and reports status: distribution by tier, delegation count and average latency.
Quickstart
bash
git clone https://github.com/AlexPEClub/Jev-Model-Router-Claude-Code-.git jev-model-router
cd jev-model-router
claudeWatch out
Grounded on a truncated README and GitHub metadata captured 2026-10-15; the README is in German and the digest cuts off in the command table, so /jev test is described only as a dry run. No license is listed in the GitHub metadata.
Questions about this entry
Short answers drawn from the linked project or page.
A hook plus skill for Claude Code that asks, for every prompt, which Claude model the task really needs. Jev from TypeSafe AI answers in under half a second per the README; if the tier is below your session model and the task is self-contained, Claude delegates it to a subagent on the cheaper model.
A UserPromptSubmit hook sends your prompt text to Jev, which answers two typed questions: the cheapest tier (haiku, sonnet, opus or fable) and the probability that the task depends on the previous conversation. The hook then prints DELEGATE or handle.
Only the text of your prompt, up to 12,000 characters. The README states no files, no code and no conversation history are sent. /jev off turns the router off entirely, after which nothing leaves the machine.
The router is fail-open: if Jev does not answer within 5 seconds, your prompt passes through unchanged. The hook never blocks your work on a routing decision.
Claude Code, jq and curl (on a Mac, brew install jq; curl is preinstalled), and an API key from console.typesafe.ai. New accounts receive starting credit according to the README.
When a prompt stays in the session, the answer starts with the tier Jev recommended and the context score, for example '-> opus (Jev: haiku, ctx 0.89)'. Delegated prompts start with '-> haiku (Jev)'. /jev status reports the tier distribution, delegation count and average latency.
Clone the repository, start Claude Code inside the cloned folder, and paste the setup prompt from the README, which copies the router files into ~/.claude and adds the hook block; then restart Claude Code and run /jev on. You add your own key to a local .env as instructed.
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
X: OpenRouter Decision Model Rankings
X: herdr + jev, an agent router
X: A group chat bot with decision room routing
More Jev resources like this
Effortlane is a local router for Codex on macOS that picks model and reasoning effort with Jev before each user turn, keeps native ChatGPT authentication, and records routing evidence locally. Shadow mode logs recommendations without changing execution.
Jev Integrations & Adapters#community#catalog-import#logicrw
A Claude Code hook that fires when you write code containing an LLM call or a hand-written classifier and injects a three-way test: plain code, a frontier model, or the System One band where Jev lives. It never blocks; the GitHub description cites a measured 149-row Jev vs Claude Haiku 4.5 comparison.
Jev Integrations & Adapters#community#catalog-import#logicrw
Local tool-routing classifier for coding agents: a gateway and MCP tool that ask Jev which tool to call next, keep your agent's own model and login, and log every decision so you can start by observing.
Jev Integrations & Adapters#community#catalog-import#logicrw
Community
felpsdev/jev-classifierTypeScript
felpsdev/jev-classifierTypeScriptA Claude Code plugin that picks the model tier and thinking effort for every turn with Jev and weighs each switch against what it costs, since moving a long conversation to another model rewrites its prompt cache and can cost more than it saves.
Jev Integrations & Adapters#community#catalog-import#logicrw
A fork of codex-with-chatgpt that adds TypeSafe Jev plus deterministic rules to decide when Codex hands planning, debugging or review to web ChatGPT and when it switches back. Smart switching is off by default, and with it off the tool behaves exactly like the original.
Jev Integrations & Adapters#community#catalog-import#logicrw
A Claude Code mod that adds $.jev to the engine interface, letting plugin hooks ask Jev for Choice, Score and Noul answers over one state via TypeSafe, OpenRouter, Vercel, Cloudflare, LiteLLM or a custom endpoint.
Jev Integrations & Adapters#community#catalog-import#logicrw
Community
chrishan17/claude-jev-modTypeScript
chrishan17/claude-jev-modTypeScriptFrom the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.