Prompt Engineering & Structured Output

Write prompts Claude can act on reliably and produce machine-validatable output. Master explicit criteria, few-shot examples, chain-of-thought, and system/role prompts; then guarantee structured output with tool_use + JSON Schema, build validation-retry loops, and scale review with multi-pass instances and the Message Batches API.