Evaluate and iteratively refine AI agent outputs through self-critique.
npx skills add https://github.com/xurenlu/marstaff --skill agentic-eval-xurenlu
Please help me install this Agent Skill. Skill: agentic-eval Source: https://github.com/xurenlu/marstaff/tree/main/skills/agentic-eval Command: npx skills add https://github.com/xurenlu/marstaff --skill agentic-eval-xurenlu
This Skill addresses the challenge of ensuring high-quality, accurate, and reliable outputs from AI agents by enabling them to evaluate and improve their own work iteratively.
Use the agentic-eval skill to refine the generated Python code for the given specification until it passes all tests.
High-intent search queries and answers about installing and using this skill.
Self-critique improves AI agent outputs by applying iterative refinement patterns where agents analyze their generated content against defined criteria to enhance quality and accuracy.
An evaluator-optimizer pipeline for LLM-as-judge systems separates generation, evaluation, and optimization roles to enable robust quality control and iterative refinement of AI outputs.
AI evaluation refines generated code to pass unit tests through test-driven code generation patterns that iteratively fix and improve outputs based on evaluation scores.
Iterative refinement requires structured output for reliable parsing of critique results and evaluation scores to ensure the agent can accurately process self-generated feedback.
Evaluating AI-generated reports for clarity and accuracy before final output uses rubric-based evaluation where the agent critiques its own work against defined criteria.
Limitations of LLM-as-judge for code quality evaluation include the dependency on structured output for reliable parsing and the need for defined criteria to drive effective self-critique loops.
alexandereiseghohiEvaluate and refine AI agent outputs through self-critique and reflection loops.
joshka0Execute JSON envelope DAGs to orchestrate multi-agent development workflows.
xabierlameiroEvaluate and refine AI agent outputs through self-critique and iterative loops.
forgivesam168Evaluate AI agent outputs using structured rubrics and JSON feedback loops.
