Claude-skill-registry best-practice-core

[Dev] Extracts and organizes best practices for a given topic into a minimal tree structure (max depth 3, max 5 children per node). Use during task planning when writing subtasks in Docs/{name}_Task.md - output is added under each subtask as a concise reference guide. Pure reasoning task with strict formatting rules: keywords/noun phrases only, no prose. (project)

install
source · Clone the upstream repo
git clone https://github.com/majiayu000/claude-skill-registry
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/majiayu000/claude-skill-registry "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/data/best-practice-core" ~/.claude/skills/majiayu000-claude-skill-registry-best-practice-core && rm -rf "$T"
manifest: skills/data/best-practice-core/SKILL.md
source content

Best Practice Core

Extracts core best practices for a topic and formats them as a minimal, keyword-focused tree.

Purpose

Extract essential best practices for implementation tasks and format them as ultra-concise tree structures for quick reference during development.

When to Use

Invocation Timing:

  • During task planning phase (before implementation)
  • When writing subtasks in
    Docs/{name}_Task.md
    documents
  • BEFORE Worker agents start implementation

Output Location:

  • Written directly under each subtask in Task document
  • Acts as quick reference for developers/agents

Output Rules (STRICT ENFORCEMENT)

Format Constraints

RuleConstraint
Max depth3 levels
Max children per node5 items
Leaf node max length12 words OR 60 characters
Node formatNoun phrases/keywords ONLY
DuplicatesMerge into single item
Tree formatMarkdown tree ONLY (no mixing)

Zero Tolerance Items

FORBIDDEN:

  • Intro paragraphs, conclusions, or commentary
  • Full sentences in leaf nodes
  • Explanatory text outside the tree
  • Mixed tree formats (ASCII + Markdown)
  • Depth > 3 or children > 5
  • Duplicate items

REQUIRED:

  • ONLY output the tree structure
  • Use noun phrases: "Error boundary setup", "State validation logic"
  • Keep leaf nodes concise: max 12 words
  • Merge similar concepts
  • Single tree format (Markdown bullets)

Example Output

Topic: Implementing User Authentication

- User Authentication
  - Security
    - Password hashing (bcrypt/argon2)
    - JWT token management
    - HTTPS-only cookies
  - Validation
    - Input sanitization
    - Email format check
    - Rate limiting
  - Error Handling
    - Failed login attempts tracking
    - Account lockout mechanism
    - Clear error messages (no data leaks)

Workflow

When invoked with a topic:

  1. Identify Core Areas: Extract 2-5 main categories for the topic
  2. Extract Best Practices: For each category, list 2-5 key practices
  3. Format as Keywords: Convert to noun phrases (no full sentences)
  4. Verify Constraints: Check depth ≤ 3, children ≤ 5, length ≤ 12 words
  5. Merge Duplicates: Combine similar/overlapping items
  6. Output Tree ONLY: No intro, no conclusion, just the tree

Integration with Task Documents

Example Task Document Integration:

## Worker1

- [ ] Implement user authentication #auth !high Deadline(2025:01:15)
  - [ ] Design login UI layout
    <!-- Best Practice Tree -->
    - Login UI
      - Layout
        - Mobile-first responsive design
        - Focus management (accessibility)
        - Password visibility toggle
      - Validation
        - Real-time feedback
        - Clear error states
      - Security
        - No password in URL params
        - Auto-logout on idle
  - [ ] Create API integration
    <!-- Best Practice Tree -->
    - Auth API Integration
      - Request Handling
        - Token refresh logic
        - Retry with exponential backoff
      - Error Handling
        - Network failure graceful degradation
        - 401/403 redirect to login
      - Security
        - Secure token storage
        - XSS/CSRF protection

Usage Notes

  • Concise over Complete: Focus on critical practices, not exhaustive lists
  • Actionable Keywords: Use phrases developers can immediately act on
  • Context-Aware: Tailor to the specific subtask context (UI vs API vs DB)
  • No Duplication: If practice applies to multiple subtasks, mention once in parent

Common Topics

Topic TypeCore Areas to Cover
UI ComponentsLayout, Accessibility, State, Events, Performance
API IntegrationRequest/Response, Error Handling, Caching, Security
Database OperationsSchema Design, Query Optimization, Transactions, Validation
State ManagementData Flow, Mutations, Side Effects, Persistence
TestingCoverage, Edge Cases, Mocking, Performance

Remember: Output ONLY the tree. No explanations.