Run structured evaluation and iterative refinement loops for AI agent outputs.
npx skills add https://github.com/miptah21/skills --skill agentic-eval-miptah21
Please help me install this Agent Skill. Skill: agentic-eval Source: https://github.com/miptah21/skills/tree/main/.agents/skills/agentic-eval Command: npx skills add https://github.com/miptah21/skills --skill agentic-eval-miptah21
Agent outputs often fail quality standards because they are produced in a single pass without structured checking, feedback, or iterative refinement.
Use the agentic-eval skill when you want your agent to generate a draft, evaluate it against explicit criteria, then refine it until it passes a score threshold.
High-intent search queries and answers about installing and using this skill.
An LLM-as-judge evaluation loop generates an agent draft, evaluates it against explicit pass/fail criteria or rubrics, and then refines the output until it passes a defined score threshold.
Automated test-driven code refinement produces code, generates tests, and iteratively refines the implementation until tests pass, using structured machine-parseable evaluation results like JSON to drive the fixes.
Self-critique reflection loops prevent endless rewriting by enforcing explicit iteration limits and convergence checks, such as score thresholds, to stop the generate-evaluate-refine cycle once criteria are met.
Yes, you can evaluate analytical writing using rubric-based scoring with weighted dimensions, requiring the evaluation results to be structured and machine-parseable so the refinement loop can apply the feedback.
Evaluator-optimizer pipelines separate generation from optimization by applying outcome-based checks, LLM-as-judge comparisons, and rubric-based scoring to control quality thresholds across code and reports.
Yes, iterative refinement workflows require structured, machine-parseable evaluation results such as JSON to ensure the evaluator feedback can be automatically parsed and applied during the refinement cycle.
selfagencyEvaluate and iteratively refine AI agent outputs using structured rubrics.
Sheldon-92Execute test suites and report failures with coverage metrics.
GPTtangEvaluate and refine AI agent outputs through self-critique and iterative loops.
ncksolEvaluate and refine AI agent outputs through structured self-critique loops.
