prompt-repetition

Prompt repetition improves LLM accuracy by providing a second pass over the input to reinforce key concepts and reduce errors in reasoning-free tasks. It targets lightweight models to enhance reliability for context, multiple-choice, or position-based tasks.

Use prompt repetition for context-rich questions, options-first multiple-choice queries, and inventory or sequence lookups. It enhances accuracy and consistency specifically in reasoning-free tasks where lightweight models might otherwise miss key context.

You should not apply prompt repetition to chain-of-thought prompts. The skill includes safety checks that automatically prevent application to chain-of-thought prompts and uses a marker to avoid duplicate application on general prompts.

For position-based or slot-based tasks, the defined repetition strategy applies the prompt 3 times. General prompts use a 2x repetition strategy to reinforce context and improve response reliability.

Yes, prompt repetition automatically applies to lightweight models like GPT-4o-mini, Claude Haiku, and Gemini Flash. It targets these specific models to improve accuracy and consistency across various reasoning-free tasks.