Skip to content

Jev Router.nvim

Neovim plugin that classifies your prompt with a single Jev choice question over the OpenRouter Decisions API, then dispatches to one of four native handlers: open a file, edit code, run a command, or answer a question.

Stars
2
Language
Lua
Published by
Community
Author
Mawfyy
Last verified
Oct 2026

Mawfyy/jev-router.nvim

User repository on GitHub

View on GitHub

An intent router for Neovim that classifies AI commands with Jev (TypeSafe's System One model) via the OpenRouter Decisions API, then routes execution to the matching handler.

Forks
1
Last pushed
Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

About this project

An intent router for Neovim that classifies AI commands with Jev (TypeSafe's System One model) via the OpenRouter Decisions API, then routes execution to the matching handler. The plugin sends the user's prompt as state, asks Jev a single choice question (intent), and maps the result to one of four routes: Neovim 0.10+ (vim.system), and curl on $PATH. git for file-candidate discovery (readfile falls back to find if absent). No external Lua dependencies are required.

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

Highlights

  • One `choice` question named `intent` maps the prompt to read_file, edit_code, run_command or general_question.
  • Edits are applied as changed hunks behind a vimdiff preview, and generated commands appear in vim.ui.input before they run.
  • Low confidence or an unavailable API drops read_file to a vim.ui.select picker instead of guessing which file you meant.
  • chat_backend, chat_stream_backend and file_provider are swappable seams, which is also how the test suite avoids network calls.

Watch out

Jev only classifies; a separate chat model writes the code, generates the command and answers the question, so a request costs two model calls through OpenRouter. Defaults are typesafe/jev-1.13 for routing and openai/gpt-4o-mini for chat, with a tiered pair used for general_question. If the model declines to return structured edits, edit_code falls back to a full-file rewrite, so the diff you approve can be much larger than expected. It needs Neovim 0.10 or newer, curl on PATH, and git for file discovery. Injected context is bounded by default at 200 files and 20,000 characters. 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

20GitHub stars
TypeScript SDK and CLI coding agent where Jev only picks options while your code runs effects, building programs as syntax trees.
Jev Agents & MCP#community#awesomejev#catalog-import
25GitHub stars
Agent-oriented playbook for using Jev as a cheap, fast decision layer that triages, classifies, scores, and gates work before expensive browser, research, retry, or subagent steps.
Jev Agents & MCP#community#awesomejev#catalog-import
13GitHub stars
Pi extension that classifies each task with Jev over OpenRouter and routes it to a Pareto-optimal model balancing measured quality, cost, and latency under a role policy.
Jev Agents & MCP#community#jevbest#catalog-import
16GitHub stars
Shell wrapper that intercepts unknown subcommands, reads the target CLI help output, and asks Jev which documented subcommand matches the intent behind the typo.
Jev Agents & MCP#community#jevbest#catalog-import
23GitHub stars
Coding-agent plugin that routes prompt classification, subagent tiering, edit rule checks, and fast Jev-backed compaction through small typed judgments.
Jev Agents & MCP#community#jevbest#catalog-import
12GitHub stars
Local OpenAI-compatible proxy that skips repeat model calls on same-intent paraphrases and retried agent turns, judged by Jev or a local System One model.
Jev Agents & MCP#community#awesomejev#catalog-import
Back to all resources

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai