What problem does it solve? Interviewers hiring for agent harness, agent infra, and agent runtime roles often lack a structured question bank covering runtime loops, tool permissions, sandboxing, sub-agent orchestration, context caching, memory externalization, trajectory evaluation, and failure recovery, leading to shallow framework-trivia interviews. ## Core Features & Use Cases - Topic Ladders: Provides progressive question ladders across nine domains including runtime loops, tool permission tiers, sub-agent orchestration, context engineering and caching, memory and skill externalization, trajectory evaluation, durable execution, human-in-the-loop confirmation, and multi-provider structured output. - Good vs Bad Question Contrasts: Contrasts weak framework-trivia questions with scenario-based questions that probe real failure cases, budgets, and verification methods. - Resume Hook Mapping: Maps resume signals (tracing, eval pipelines, cache hit rates, Claude Code / Codex / Cursor usage) to targeted follow-up probes. - Use Case: When a job description mentions agent harness, agentic coding, or eval pipelines, load this Skill to generate calibrated interview questions with expected strong and weak answer signals for campus or experienced hires. ## Quick Start Load this skill and generate interview questions for a candidate whose resume mentions building an agent runtime with tool calling and trajectory evaluation.
