Generate 20 spreadsheet-ready evaluation test cases for AI features.
npx skills add https://github.com/breethomas/bette-think --skill start-evals
Please help me install this Agent Skill. Skill: start-evals Source: https://github.com/breethomas/bette-think/tree/main/plugins/bette-think/skills/start-evals Command: npx skills add https://github.com/breethomas/bette-think --skill start-evals
Reduce overengineering and uncertainty when validating AI features by providing a concise, repeatable evaluation starter that emphasizes error analysis over premature automation. The approach helps teams quickly surface failure modes with a small, high-value set of tests so they can iterate product and model improvements faster.
Type /start-evals "AI product recommendations" to generate 20 spreadsheet-ready test cases, pass/fail criteria, and an actionable evaluation workflow.
High-intent search queries and answers about installing and using this skill.
AI evaluation test cases are structured inputs designed to validate specific AI features like recommendation engines or summarization tools. This approach generates 20 spreadsheet-ready tests, including 15 happy-path and 5 edge-case inputs, to surface failure modes with pass/fail criteria during initial QA.
The best way to start QA for an AI product recommendation system is applying a focused 20-test starter kit formatted for spreadsheets. This method emphasizes error analysis over premature automation, providing a Week 1 workflow to surface failure modes and iterate model improvements within 30 to 90 minutes.
Yes, spreadsheet-driven QA works for RAG pipeline validation by providing 20 structured test cases with pass/fail criteria. This approach fits early-stage retrieval-augmented generation systems, enabling rapid error analysis and product iteration without requiring complex automated testing infrastructure.
You scale AI evaluation tests from 20 to 200+ cases by following provided graduation guidance to incrementally expand test coverage. The workflow supports this scale-up by introducing options to build an LLM-as-judge for automated scoring and optional Linear project creation for tracking.
No, an automated testing framework is not required to initially validate customer support assistants. A pragmatic spreadsheet-driven workflow using 20 targeted test cases helps you quickly identify failure modes and perform error analysis before investing in premature automation or complex infrastructure.
jshsakuraDocument regression tests for prompt, model, or workflow changes.
winhokExecute API test specs from YAML, JSON, Excel, or CSV and report pass/fail results.
majesticlabs-devGenerate referral program messaging templates for emails, landing pages, social shares, and thank-you notes.
seo-jinseokSimulate user personas and score Regulation Q&A responses with LLM-as-Judge.
