prompt refiner pro

The problem

Vague instructions lead to inconsistent AI outputs, hallucinations, and ignored constraints. Writing production-grade prompts is a trial-and-error process that wastes developer time and token costs.

What it does

• Audits existing prompts against 12 failure patterns including role clarity and persona drift.
• Rewrites weak prompts into structured, production-grade instructions with explicit roles and constraints.
• Generates a scored audit report to identify high-risk areas in your current implementation.
• Provides a 5-point testing protocol to validate the new prompt against edge cases and adversarial inputs.
• Includes a before/after comparison to track improvements in output consistency and token efficiency.

Frameworks & tools

Claude Code, Cursor, Claude.ai, and OpenAI Codex.

Why this beats prompting it yourself

Instead of guessing why a prompt fails, you get a systematic diagnostic report. This skill replaces intuition with a repeatable framework that ensures every instruction follows structured engineering principles like few-shot examples and output schemas.

Use cases

• Converting a casual chat prompt into a formal system prompt for an API.
• Fixing prompts that consistently ignore formatting rules or negative constraints.
• Reducing token waste by trimming redundant context from legacy prompts.
• Hardening agent instructions to prevent persona drift during long conversations.

Known limitations

Does not fix hardware limitations, model-specific knowledge gaps, or retrieval issues unrelated to prompting.