orchestrate-prompt-engineeringlisted

# Orchestrate: Prompt Engineering as a Reviewed Artifact **Direct evidence**: HackerRank's guidance states this directly — *"Write prompts with same care as code—specify allowed outputs, required evidence, format requirements."* This sits under "Code Quality," meaning prompts are implicitly graded as part of that 30%, not treated as separate from "real" engineering. ## What "same care as code" concretely means A prompt written with code-level rigor has the same properties good code has: - **Explicit allowed outputs.** Not "classify the ticket" but "classify as exactly one of: `product_issue`, `feature_request`, `bug`, `invalid` — no other values." This is the prompt-side half of `orchestrate-schema-guardrails`'s validation-side guardrail; the two should agree exactly. - **Required evidence, stated as a constraint, not a hope.** "Ground your response only in the provided corpus documents; if no relevant document exists, say so explicitly rather than answering from general knowledge" — directly enforcing the "must use only the provided support corpus" hard constraint and preventing hallucinated policy citations, which the starter repo names as something to avoid. - **Format requirements stated precisely.** If you need JSON, specify the exact keys and types, not "respond in JSON." If you need a justification under a certain length, say so — vague format instructions produce vague, inconsistently-parseable output. ## Treat prompts as versioned, readable files — not inline st