21 October 2026
اختبر الـ plugin عندك بـ claude plugin eval وشوف إذا كان ينفع فعلاًTest Your Claude Code Plugin with claude plugin eval
You will test your Claude Code plugin with real eval cases and see in numbers whether it improves the result, or whether Claude would have passed without it.
Who this guide is for
Anyone building a plugin or skill for Claude Code who wants proof it works before sharing it with a team.
Key terms
- Plugin eval
- A shell command, claude plugin eval, that runs your plugin against a suite of test cases and scores each one.
- Case
- A folder inside evals with a realistic request, written the way a user would type it, plus one or more graders.
- Grader
- A pass or fail check on what Claude produced. A regex on the reply, whether a tool was called, or a second model judging against a rubric you write.
- Baseline
- The same case run with no plugin. The gap between the two scores is Δ, which is what the plugin actually added.
The steps
Step 1
Check the requirements
You need Claude Code 2.1.269 or later. Run claude --version to check and claude update to upgrade. If git is installed it must be 2.31 or later.
Every run and every judge grader is a real model call on your account, counted against your plan or API bill. Start with one or two cases.
PromptCheck versionsclaude --version git --version
Tip: Open the terminal in your plugin's root folder, the one with plugin.json.
7 more steps wait for you after access
With the ready prompts and the takeaway
Bought this before? Restore your access
Your access is tied to your email. Enter it here and the content opens right away, any time, on any device
Most value
Start building with me now
We sit together, break down your problem, and I build the AI solution with you step by step
- A private session with you
- Real building during the session
- Results you take with you when we finish
100% free
A free consultation about your problem
