Oro Miner Prompt Optimization Yields Score Gains

A miner tested 15 prompt iterations and deployed one successful change that improved parsing of multi-product queries, raising the miner's score from 0.4286 to 0.8571. The winning adjustment replaced vague numbered phrasing with explicit separator patterns like "both X and Y" to ensure each product gets its own entry. Product, shop, and voucher categories each score 0.3, indicating room for further optimization. Updated code is live on GitHub.

• •Score improved from 0.4286 to 0.8571 via prompt refinement
• •Change: explicit separator patterns for multi-product query parsing
• •All three category scores at 0.3; additional gains possible
• •Updated miner code deployed to production