senior-prompt-engineer

RAG evaluation involves analyzing token efficiency, faithfulness, and safety using structured workflows. The rag_evaluator script loads prompts, runs analyses, and generates validation reports to measure retrieval accuracy and output reliability.

Prompt engineering patterns like Zero-shot, Few-shot, Chain-of-Thought, ReAct, and Tree of Thoughts reduce hallucinations by structuring reasoning steps. The prompt_optimizer script analyzes and optimizes prompts to ensure robust, cost-efficient outputs.

Agent orchestration primitives coordinate multi-tool and multi-agent workflows using reference architectures. The agent_orchestrator script manages tool integration across Claude Code, Codex, and OpenClaw contexts to execute complex agent design frameworks reliably.

The Python scripts integrate into existing tooling to load prompts and run analyses. Reusable templates and reference documentation accompany the scripts, allowing seamless incorporation into established AI evaluation and agent design pipelines.

Structured output patterns are needed when generating validation reports or coordinating multi-agent workflows. They ensure robust, cost-efficient outputs by enforcing format constraints, which is critical for agent orchestration and end-to-end prompt evaluation.

Optimizing complex prompt workflows requires combining prompt engineering patterns, RAG evaluation, and agent orchestration. Running the prompt_optimizer, rag_evaluator, and agent_orchestrator scripts sequentially analyzes, optimizes, and validates the entire workflow.