Categorize engine test results into core gaps, extension opportunities, and lessons learned.
npx skills add https://github.com/rayanino/kr --skill kr-evaluate
Please help me install this Agent Skill. Skill: kr-evaluate Source: https://github.com/rayanino/kr/tree/main/skills/kr-evaluate Command: npx skills add https://github.com/rayanino/kr --skill kr-evaluate
This Skill streamlines the review of engine test results, ensuring thorough categorization of findings and clear distinction between critical issues and potential improvements.
Use the kr-evaluate skill to review the test output for the 'arabic-names' engine.
High-intent search queries and answers about installing and using this skill.
To review automated engine test results, categorize each finding into core gaps, extension opportunities, or lessons learned. You must distinguish between code bugs, prompt issues, data problems, and evaluator noise to ensure thorough quality assurance.
Evaluating LLM-worker quality assessment outputs involves categorizing findings into core gaps or extension opportunities. You need to distinguish prompt issues and evaluator noise from fundamental code bugs to determine overall engine reliability.
Distinguishing code bugs from prompt issues in LLM evaluator reviews requires categorizing test findings. You separate fundamental flaws requiring immediate fixes from data problems or evaluator noise to produce a comprehensive aggregate assessment.
Yes, automated test categorization supports Arabic content and LLM-generated outputs. It facilitates targeted owner spot-check reviews with domain experts, categorizing findings into core gaps, extension opportunities, and lessons learned for quality assurance.
You need to categorize engine test findings into core gaps and extension opportunities after running a batch of tests on a new language engine. This process identifies fundamental flaws requiring fixes and areas for future enhancement to determine readiness.
Inconsistent pass/fail test results during an LLM-evaluator review often stem from evaluator noise, prompt issues, or data problems. Categorizing these findings helps distinguish evaluator noise from fundamental code bugs to reach an overall verdict.
brainbrewlabsRun test suites and generate pass/fail verdicts with coverage reports.
GuidoBergmanConduct post-epic reviews to assess success and extract lessons.
doruciocleaGenerate SEO-optimized cybersecurity articles about AI-generated code vulnerabilities.
imbue-aiAdd new issue types to review-rule documents for React, Sculptor frontend, and integration tests.
