workflow-from-chats

To extract reusable workflow preferences from Cursor chats, this Skill mines recent transcripts for explicit preferences and corrections, deriving preference atoms with confidence ratings to generate actionable rules or skills.

Turning AI coding chat history into durable rules and skills requires inventorying parent and subagent transcript evidence, deriving preference atoms with decision rules, and selecting appropriate artifact types like new skills or rule edits.

This workflow extraction process applies safety filtering to ensure private chat content, transcript paths, and credentials are never exposed, while still successfully deriving durable working preferences from your chat history.

You can personalize agent behavior by mining feedback from your last 7 days of Cursor conversations, extracting durable practices, clustering them by workflow shape, and outputting personalized workflow direction or skill edits.

The default time window for mining feedback from chat transcripts is the last 7 days, ensuring extracted workflow guidance reflects your most recent working preferences and current agent interaction patterns.