Hi Product Hunt, I built jev-cli because I wanted to use TypeSafe Jev from shell scripts and coding agents without writing a new API wrapper for every project. The package installs two commands: - `jev` for shell scripts, JSON pipelines, and application automation - `jev-mcp` for agents and other MCP-compatible hosts The CLI supports Jev's three judgment types: `noul`, `choice`, and `score`. It can also send multiple questions in one batch request. Output is compact JSON by default, while diagnostics go to stderr with documented exit codes. Install it from PyPI with: uv tool install jev-cli The official TypeSafe API is the default provider. You can also use Jev through Vercel AI Gateway, OpenRouter, or a compatible custom endpoint. jev-cli also includes an optional Agent Skill for coding agents that work better with command-line tools: jev install-skills This is an independent community project. It is not affiliated with, maintained by, or endorsed by TypeSafe AI. I would be interested to hear where small semantic judgments could replace larger generative-model calls in your software.

jev-cli

TypeSafe Jev for shell scripts and AI agents

TypeSafe Jev for shell scripts and AI agents