What does plugin-eval do?

It lets your agent score and rank other plugins by quality using an automated three-level review system, producing Elo rankings from the results.

What are the three layers?

The item describes a three-layer evaluation framework but does not specify what each layer contains.

Can it evaluate plugins on its own?

Yes, the review system is automated, so the agent can run evaluations without manual inspection of each plugin.

plugin-eval

PackDev tools

plugin-eval is a plugin that gives an AI agent a three-layer quality evaluation framework for Claude Code plugins. It runs an automated review at three levels and produces Elo rankings, so the agent can score and compare other plugins by quality instead of relying on manual inspection.

Unavailable. Delivery for this kind is on the roadmap — not serving yet.

Have an agent setup that supports Claude Code plugins.

What your AI can do with it

Runs a three-layer automated review of Claude Code plugins
Scores plugins for quality
Ranks plugins against each other using Elo
Lets the agent evaluate other plugins automatically

Getting started

Have an agent setup that supports Claude Code plugins.
Install the plugin-eval plugin.
Point it at the plugins you want evaluated.
Run the evaluation to get scores and Elo rankings.

Signals

GitHub stars: 40k
Forks: 4k
Last commit: Sep 2026

Questions

What does plugin-eval do?: It lets your agent score and rank other plugins by quality using an automated three-level review system, producing Elo rankings from the results.
What are the three layers?: The item describes a three-layer evaluation framework but does not specify what each layer contains.
Can it evaluate plugins on its own?: Yes, the review system is automated, so the agent can run evaluations without manual inspection of each plugin.

Advanced

Item type: plugin
Key: wshobson-agents-plugin-eval
Source: github.com/wshobson/agents

github.com/wshobson/agents