Evaluate AI agent performance using multi-dimensional rubrics and LLM-as-judge.
npx skills add https://github.com/GobbyAI/gobby --skill eval-gobbyai
Please help me install this Agent Skill. Skill: eval Source: https://github.com/GobbyAI/gobby/tree/main/src/gobby/install/shared/skills/eval Command: npx skills add https://github.com/GobbyAI/gobby --skill eval-gobbyai
The eval skill enables systematic evaluation of AI agents' performance and overall quality, helping teams quantify improvements and detect regressions.
Activate the eval skill by prompting: "Evaluate current agent performance against the latest test suite" and review the resulting rubric scores and recommendations.
High-intent search queries and answers about installing and using this skill.
To evaluate AI agent performance systematically, build test frameworks with multi-dimensional rubrics and apply LLM-as-judge methodologies. This quantifies improvements and detects regressions across diverse use cases.
LLM-as-judge is an evaluation mechanism using language models to scale assessments of agent outputs. You need it when enforcing continuous quality gates for agent pipelines requires automated, multi-dimensional scoring.
Set up quality gates for agent pipelines by defining evaluation rubrics and integrating LLM-as-judge checks to enforce continuous quality control and detect regressions before deployment.
Yes, multi-dimensional evaluation rubrics apply to diverse agent use cases. The framework specifies practical rubric formats and activation guidance to support repeatable assessments across various scenarios.
The best way to detect regressions in AI agent quality is to enforce continuous quality gates using test suites and LLM-as-judge. Reviewing resulting rubric scores highlights performance drops and actionable improvements.
Sir-chawakornBuilds an LLM evaluation harness using golden datasets and code-based scorers.
erichowensLog and score skill usage quality with a defined invocation data schema.
FVossebeldEvaluate agent outputs with weighted rubric criteria and per-dimension scores.
NVIDIA-NeMoStandardize LLM agent and RAG pipeline evaluation with composable metrics and reproducible benchmarks.
