evaluation

Evaluate agent performance across multiple dimensions using multi-dimensional rubrics.

Updated Jun 12, 2026
One-click install
npx skills add https://github.com/Kushal9889/claude-plugins --skill evaluation-kushal9889
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: evaluation
Source: https://github.com/Kushal9889/claude-plugins/tree/main/context-engineering/skills/evaluation
Command: npx skills add https://github.com/Kushal9889/claude-plugins --skill evaluation-kushal9889

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires pypdf, pdfplumber, pdf2image, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill addresses the challenge of evaluating agent performance systematically, ensuring the quality of context engineering choices and the effectiveness of agent systems.

Core Features & Use Cases

  • Systematic Evaluation: Provides a framework for evaluating agent performance across various dimensions.
  • Multi-Dimensional Rubrics: Offers multi-dimensional rubrics for factual accuracy, completeness, citation accuracy, source quality, and tool efficiency.
  • Continuous Monitoring: Integrates with production systems for ongoing evaluation and monitoring.

Quick Start

Run the evaluation for the 'agent-systems' test set using the 'evaluate' command.

Frequently Asked Questions about evaluation

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I evaluate agent performance across multiple dimensions systematically?

You evaluate agent performance by applying a multi-dimensional rubric that measures factual accuracy, completeness, citation accuracy, source quality, and tool efficiency to ensure your agent systems function effectively.

What is context engineering quality assurance for AI agents?

Context engineering quality assurance is the systematic evaluation of agent systems using multi-dimensional rubrics to validate factual accuracy, source quality, and tool efficiency, ensuring continuous monitoring of production environments.

Can I use this agent evaluation framework for continuous monitoring in production?

Yes, the agent evaluation framework integrates with production systems for continuous monitoring, allowing you to run ongoing evaluations against test sets to maintain context engineering quality gates.

Do I need Python libraries to run agent performance testing and test set management?

Yes, you need specific Python libraries including pypdf, pdfplumber, and pdf2image to manage test sets and execute the scripts required for comprehensive agent performance evaluation.

How do I run an evaluation for an agent systems test set?

You run an evaluation for the 'agent-systems' test set by executing the 'evaluate' command, which triggers the multi-dimensional rubrics to analyze factual accuracy and tool efficiency.

What are the limitations of using multi-dimensional rubrics for agent testing?

The multi-dimensional rubric approach for agent testing requires Python dependencies like pdfplumber and pypdf, meaning evaluation is limited to environments where these specific test set management libraries are installed.