The Voice Learn skill bridges the gap between AI-generated drafts and a user's final polished work by systematically identifying patterns in manual edits. By comparing raw output, reviewed versions, and final post-edit copies, it detects nuances in tone, structure, and register. It maintains a pattern 'accumulator' to ensure only consistent preferences become permanent rules, helping to eliminate 'voice drift' and reduce the need for repetitive manual corrections. This skill is essential for users who require high-precision brand alignment or a distinct personal writing voice in their documentation and content.
主要功能
01Three-stage snapshot comparison (Pre-review, Post-review, Post-edit)
02307 GitHub stars
03Automated register and craft-rule update proposals based on evidence
04Staleness management to prune outdated or inconsistent voice traits automatically
05Pattern accumulation logic to prevent rule bloat from one-off edits
06Contradiction detection to flag and resolve conflicting style preferences
使用场景
01Ensuring long-term consistency across large-scale content and documentation projects
02Training Claude to match a specific corporate brand or technical documentation tone
03Reducing repetitive post-editing tasks by automating style adjustments