continuous-learning

After completing a complex task, prompt the system to learn and reflect. It collects your context, decisions, and patterns, then generates a standardized skill file in ~/.claude/skills/ with metadata and a concise summary you can reuse across future tasks.

Create a skill when you encounter novel patterns, solve complex problems, or complete tasks you'll repeat. The system identifies generalizable insights and packages them as reusable skills that improve your workflow over time.

Generated skill files include metadata, a structured description of the pattern or technique, and a concise summary. Files are saved under .claude/skills and ready to apply to similar future tasks without manual setup.

This works after any complex task where you want to capture and generalize insights. It's domain-agnostic—applicable to software engineering, data analysis, writing, or any workflow where patterns repeat and reflection adds value.

Reflection analyzes outcomes and decisions from a single task. Saving as a skill extracts generalizable patterns and packages them into reusable Claude skills that apply across multiple similar tasks and contexts.

The system applies quality gates during skill extraction to ensure learnings are specific, generalizable, and repeatable. It filters out task-specific noise and surfaces only patterns applicable to future similar work.