What problem does it solve? Writing effective LLM prompts and agent workflows is often trial-and-error, leading to wasted tokens, ambiguous outputs, and unvalidated agent designs. This Skill provides static analysis tools and reference frameworks to systematically improve prompt quality, measure RAG retrieval performance, and validate agent architectures before deployment. ## Core Features & Use Cases - Prompt Optimizer: Analyzes prompts for token count, cost estimation, clarity scores, ambiguity issues, and redundant content, then generates optimized versions and extracts few-shot examples. - RAG Evaluator: Measures context relevance, retrieval precision, answer faithfulness, and groundedness against question/context datasets. - Agent Orchestrator: Validates agent configurations (ReAct, Plan-Execute, tool-use patterns), visualizes workflows as ASCII or Mermaid diagrams, and estimates per-run token costs. - Use Case: Before shipping a customer-support agent, run its YAML config through the orchestrator to catch missing tool configs and infinite-loop risks, then optimize the system prompt to cut token costs by removing redundant instructions. ## Quick Start Ask the assistant to analyze your prompt file with the prompt optimizer script to get a clarity score, token count, and concrete improvement suggestions.
