Unofficial Jev plugin for coding agents: best practices, an API reference, and 150+ community projects. Evals included.

I'm coming up for air from the bottomless evals ocean. I made a plugin that helps your harness of choice work Jev into your projects more holistically: best practices, anti-patterns, an API reference, and links to 150+ community projects grouped by design pattern, with a code sketch for each pattern. I built it around progressive disclosure, so the agent only loads what the task needs and it doesn't eat your context. https://github.com/aaddrick/building-with-typesafe-jev The eval work is in the repo. I ran six Jev coding tasks 10 times each with no plugin, with TypeSafe's official plugin, and with this one. Judgment calls went to three judges from three providers (Claude Opus, GPT-6 Sol, Kimi K3), and the majority decided, so no model family grades its own work. Score is the share of checks passed: No plugin: 0.65 Official plugin: 0.77 This plugin: 0.96