Проектирование процесса отладки AI-систем

★ 6.9 · marketing

ai-debugging-workflow is a Claude Code skill that designs and documents an AI debugging workflow with a structured process, quality checks, and system integration guidance. It follows four steps: reading project context from memory.md and knowledge-base.md, selecting the most suitable framework from options like AI Readiness Assessment, Human-in-the-Loop Design, RAG Architecture, and Agent Orchestration Patterns, generating specific actionable recommendations, and validating the output against a built-in checklist. The resulting document covers an executive summary, an implementation table with owners and timelines, key metrics (Time Saved Per Task, Automation Rate, Error Reduction %, Cost Per AI Operation, User Adoption Rate), and a risk mitigation matrix. It suits marketing and product teams rolling out AI automation who need a measurable plan with clear KPIs rather than generic advice.