What problem does it solve? AI troubleshooting responses often lack transparency about how reliable their conclusions are, leaving engineers unsure whether to trust a diagnosis or escalate. This Skill enforces a mandatory, standardized confidence assessment on every response so users can gauge reliability at a glance. ## Core Features & Use Cases - Three-Dimension Scoring: Rates Data Source (40%), Query Accuracy (35%), and Diagnosis (25%) with 🟢/🟡/🔴 icons and computes a weighted overall percentage. - Auto-Scoring Triggers: Automatically adjusts scores based on concrete events such as KQL query row counts, MCP tool failures, schema verification, and speculative language. - Special-Case Handling: Provides scoring rules for configuration guides, code reviews, dashboard links, and multi-step investigations. - Use Case: After running an azuremcp Kusto query that returns BGP flap data for an ExpressRoute circuit, the agent appends a confidence table showing 🟢 across all dimensions with a 92% overall score, signaling the findings are safe to use in case notes. ## Quick Start Ask the agent to troubleshoot an Azure networking issue and include the confidence score table assessing data source, query accuracy, and diagnosis for its findings.
