What problem does it solve? AI-assisted sessions degrade when the model repeats mistakes, hallucinates files, or drifts from your goal, and users often notice too late. This Skill detects collaboration quality drops early and prompts structured re-anchoring before frustration escalates. ## Core Features & Use Cases - Signal Detection: Analyzes user messages for repeated corrections, hallucination reports, context confusion, goal drift, and frustration patterns using regex-based heuristics. - Severity Classification: Categorizes issues as low, medium, high, or critical and generates a diagnostic report with evidence and suggestions. - Re-anchoring Templates: Provides goal, technical, and communication re-alignment templates to confirm understanding before continuing. - Use Case: After correcting the AI twice about plural REST endpoint naming, the Skill triggers a quality check, acknowledges the correction pattern, and confirms the convention before generating more endpoints. ## Quick Start Ask the AI to run a quality check on the recent conversation because something seems off with its responses.
