master learn

Maintaining consistency in a growing codebase is difficult, especially when multiple developers or AI agents are contributing. This skill solves the problem of "architectural drift" by automatically discovering how you actually write code, documenting those patterns, and preparing them for enforcement.

What it does

• Pattern discovery scans your repository to identify specific ways you handle data fetching, error handling, naming, and folder structure.
• Statistical analysis counts occurrences of coding styles to distinguish between established conventions, split decisions, and outliers.
• Project mapping analyzes your directory tree to determine exactly where new features, components, hooks, and types should be placed.
• Conflict detection identifies areas where the current code disagrees with previously established rules in your grand-master configuration.
• Interactive decisioning prompts you to choose a standard when it finds two or more competing ways of doing the same thing.

How it works

1. Initialize by running /master-learn to trigger the stack identification and folder mapping sequence.
2. Review analysis as the agent crawls your source files to categorize coding patterns based on frequency and consistency.
3. Resolve splits by answering simple questions when the agent finds conflicting patterns (e.g., inline fetching vs. custom hooks).
4. Inspect draft by checking .claude/patterns/draft.md where the agent records candidates, evidence, and golden file examples.

Frameworks & tools

Works with any stack including React, Next.js, Vue, Express, or Django. It parses standard configuration files like package.json, tsconfig.json, requirements.txt, and go.mod to adapt its learning logic to your specific ecosystem.

Why this beats prompting it yourself

Manual prompting for architectural consistency requires you to remember every convention you have ever used. This skill automates the audit process by looking at your actual files, providing quantitative evidence for every rule it proposes rather than guessing based on generic best practices.

Use cases

• Onboarding agents to a new repository so they don't introduce foreign coding styles or folder structures.
• Standardizing a legacy project by identifying the most common patterns and marking the rest as technical debt.
• Establishing defaults for a brand new project by approving a starter set of sensible rules for your chosen framework.
• Auditing consistency before a major refactor to see which patterns are actually being followed by the team.

Known limitations

Requires an existing codebase with at least 2 files per category to form candidates. It cannot edit code directly and only writes to the draft patterns file for your review.