Quality Assurance
AGENTS.md · 6 documents
Agent Eval - Complete UI Flow Visualization
This document shows WHERE every feature appears in the application, following the user journey from start to finish.
17. Evaluation Frameworks
title: 17. Evaluation Frameworks
Lint Compliance Report
All critical lint errors have been fixed. The codebase now passes flake8 checks with NVDA-specific configuration.
Bots Against Alignment -- Rules
Bots Against Alignment is a multiplayer party game (think Cards Against Humanity, but with AI). Each player builds an AI "bot" with a custom personality, then competes to produce the response that best satisfies a shared AI judge called **the Aligner**. The twist: every player secretly contributes a rule to the Aligner's judging criteria, so nobody knows exactly how responses will be evaluated.
verifier
description: Validates completed work. Use after tasks are marked done to confirm implementations are functional.
Lesson 11 - Evals (Regression Testing for Agents)
**"How do I know if my agent still works after I change something?"**