Learn-taste skill for Claude Code helps agents avoid repeated design mistakes

Interface designer and developer Emil Kowalski recommends installing the /learn-taste skill for Claude Code.
The skill collects every design correction you make during a session and converts each one into a strict rule with an explanation of why it matters. As a result, the next agent does not repeat the same mistakes.
This is especially useful if you frequently have to fix an agent's work. It can save a noticeable amount of time.
To install it, run:
npx skills add emilkowalski/skills --skill learn-taste
At the end of a session, run:
/learn-taste Analyze my last three sessions as well. I keep fixing the same mistakes.
This command asks the skill to review three previous sessions in which you repeatedly corrected the same errors.
Why it matters
AnalysisIt turns temporary manual fixes into durable instructions for AI agents, reducing repeated errors and improving consistency in AI-assisted design sessions.
Discuss in community
Share your questions and insights with developers
Related in Ecosystem
Practical guides, core terminology, and agent skills
Claude Code desktop app adds custom themes via mods
Claude Code’s desktop app now lets users create custom themes through mods, change colors, add animated backgrounds, and customize the interface.
Claude Code Projects: How a Single AI Orchestrates a Team of Autonomous Agents
A deep dive into Claude Code Projects: Coordinator and Threads architecture, parallel branches, shared memory via MEMORY.md, and cloud agent management.
How to Build Custom Skills for Claude Code: Architecture and Best Practices
A comprehensive engineering guide to creating custom Claude Code skills: folder structure and SKILL.md, progressive disclosure, skills vs. hooks trade-offs, and evaluation-driven workflow development.
Claude Code
The official terminal agent from Anthropic, operating directly in the command line via Claude 3.7 Sonnet with native support for Bash, Git, file systems, and the MCP protocol.