Self-Evolving Skill – empirical results from a 5-round experiment
Five-gate validation framework prevents skill knowledge drift, but experiments limited to one domain.
A design pattern for Claude Code Skills that improve through use — more accurate, more efficient, never bloated. | 越用越准、越用越快、但不臃肿的 Skill 设计模式
Claude Code Skill pattern paper—interesting theory, but unclear if it ships as a usable tool today.
Claude Code users building domain-specific skills, particularly for database work and business systems
Five-gate validation framework prevents skill knowledge drift, but experiments limited to one domain.
Recursive repair loops improve skills automatically, unlike static Claude Code defaults.
Yet another Claude skill for UI consistency when Cursor rules already do this.
Design extraction API—clever, but LLM-as-designer for Figma already exists, and extraction is trivial.
Curated markdown collection when awesome lists already exist everywhere.
It watches your Claude sessions, captures corrections and repeated flows, and produces structured observation logs plus candidate SKILL.md drafts so your skill library can grow without manual audits. The pragmatic split of writing to the filesystem in Cowork versus producing a handoff doc for the web shows an awareness of platform limits. Clever idea, but the README promises methodology without live examples, metrics, or privacy details, so it's interesting on paper more than battle-tested.