What problem does it solve? Prompts often underperform because of vague instructions, missing output contracts, bloated schemas, or misapplied techniques, and teams waste effort on advanced patterns when simpler fixes would work. This Skill diagnoses the actual failure mode and recommends the smallest effective prompt-engineering technique before rewriting. ## Core Features & Use Cases - Diagnosis-first prompt improvement: Distinguishes prompt problems from retrieval, tooling, or requirements issues before recommending changes. - Technique selection guidance: Compares grounding, structured output, few-shot, chaining, reasoning patterns, RAG variants, and parameter tuning with explicit tradeoffs. - Token-budget optimization: Moves large schemas into referenced JSON Schema files, formalizes workflows as Mermaid or BPMN artifacts, and applies selective interspersed repetition. - Use Case: A team has a prompt that pastes a long JSON schema and a multi-step approval workflow inline. The Skill recommends externalizing the schema, referencing a Mermaid workflow file, and produces a compact rewritten markdown prompt. ## Quick Start Ask the agent to review your markdown prompt file, explain which prompt-engineering technique fits best, and provide a rewritten version with an explicit output contract.
