Prompts Library

Curated prompts that have proven useful in production work. Each entry includes the prompt itself, a short description, recommended models, and a tag list. Submit a prompt if you've built one worth sharing.

By category

• Classification 1 prompt
• Coding 4 prompts
• Extraction 1 prompt
• Rag 1 prompt
• Writing 3 prompts

All prompts

• Structured Data Extraction to JSON Extracts specific fields from unstructured text into a strict JSON schema. Reliable across providers when JSON mode is on.
  extraction ▲ 52
• Code Review — Specific, Severity-Graded Reviews code with concrete fixes per finding, grouped by severity. Avoids generic "consider adding tests" suggestions.
  coding ▲ 47
• Debug a Production Error Walks through a stack trace systematically: what failed, why, and the smallest change to fix it without introducing risk.
  coding ▲ 44
• Rewrite for Clarity, Keep Meaning Rewrites a paragraph to be clearer and tighter without changing the underlying claims. Preserves voice.
  writing ▲ 41
• RAG Answer — Citations Required Answers a question from provided context with inline citations. Refuses when the context doesn't support an answer.
  rag ▲ 38
• Write Unit Tests — Cover Edge Cases Generates unit tests in {{framework}} that cover the happy path, edge cases, and error conditions. Tests are runnable, not pseudocode.
  coding ▲ 35
• Intent Classification with Confidence Routes a user message to one of N categories with a confidence score. Returns "unclear" rather than guessing when uncertain.
  classification ▲ 31
• Meeting Notes — Decisions and Action Items Summarizes meeting transcript into decisions, action items (with owners), and unresolved questions. Skips small talk.
  writing ▲ 29
• Explain a SQL Query Translates a complex SQL query into plain English step by step. Notes performance characteristics and missing indexes.
  coding ▲ 26
• Technical Translation — Preserve Code and Terms Translates technical content while preserving code blocks, API names, version numbers, and untranslatable terms.
  writing ▲ 18

What makes a prompt worth saving

The prompts here aren't generic "you are a helpful assistant" starters. Each one targets a specific production use case — code review, structured extraction, RAG with grounding, classification — and includes the constraints that make it actually work at scale: explicit output schema, refusal clauses, severity grouping, citation requirements.

Submit a prompt if it meets the same bar. Include what model you used it with, what the input format looks like, and what failure modes it solves.