Prompt Evaluation Framework

Help me build an evaluation framework for a prompt I'm using in production (or planning to).

**The prompt I want to evaluate:**
[Paste your prompt here]

**What this prompt is supposed to produce:**
[Describe the ideal output — format, content, tone, accuracy]

**How critical is quality here?**
[e.g. customer-facing copy, internal tool, high-stakes decisions, low-stakes drafts]

**What "good" looks like:**
[Any specific criteria — e.g. tone must be professional, must include 3 bullet points, must not hallucinate facts]

**What "bad" looks like:**
[Common failure modes you've already seen or are worried about]

Build me a complete evaluation framework including:
1. A scoring rubric with 4–6 criteria, each rated 1–5 with clear descriptors
2. A set of 5 test inputs designed to stress-test the prompt (including edge cases)
3. A grading template I can fill in for each output
4. A pass/fail threshold recommendation
5. Guidance on when to iterate vs. when to ship