claude-blog-sources

What this skill does

# Claude Blog Sources

## When to Use This Skill

| Use this skill when... | Use alternative when... |
|------------------------|-------------------------|
| Researching Claude Code features and best practices | You already know the answer from existing documentation |
| Need latest updates on CLAUDE.md, @imports, memory hierarchy | Implementing known patterns without research needed |
| Staying current with Claude Code improvements | Just need to read local project documentation |
| Cross-referencing blog insights with official docs | Need general programming help unrelated to Claude |

## Overview

The Anthropic Claude Blog (https://www.claude.com/blog/) publishes official guidance on Claude Code features, usage patterns, and best practices. This skill provides structured access to blog content for staying current with Claude capabilities.

## Primary Sources

### Official Documentation (preferred for technical details)

**Claude Code Docs**: https://code.claude.com/docs/en/

| Topic | URL | Description |
|-------|-----|-------------|
| Memory management | /docs/en/memory | CLAUDE.md, auto memory, @imports, rules hierarchy |
| Configuration | /docs/en/configuration | Settings, permissions, MCP servers |
| Hooks | /docs/en/hooks | Pre/post tool use hooks |
| Claude Code overview | /docs/en/overview | Getting started, features |

Use official docs as the authoritative reference. Blog posts supplement with patterns and examples.

### Blog

**Main Blog**: https://www.claude.com/blog/

Note: May redirect to https://website.claude.com/blog - follow redirects automatically.

## Key Articles for Claude Code

### Essential Reading

| Article | URL | Topic |
|---------|-----|-------|
| Using CLAUDE.md Files | /blog/using-claude-md-files | CLAUDE.md structure and best practices |
| How Anthropic Teams Use Claude Code | /blog/how-anthropic-teams-use-claude-code | Internal usage patterns and workflows |
| Claude Code on the Web | /blog/claude-code-on-the-web | Web-based features and capabilities |

### Article Relevance Categories

**High Relevance** (always check):
- Articles mentioning "CLAUDE.md", "Claude Code", "Skills", "Agents"
- Articles about coding workflows, development practices
- Product announcements for Claude Code features

**Medium Relevance** (check if applicable):
- Articles about prompt engineering, context management
- Enterprise AI deployment patterns
- Integration guides (Slack, GitHub, etc.)

## Research Workflow

### When User Asks About Claude Code Features

1. **Check Recent Articles First**
   ```
   WebSearch: site:claude.com/blog {feature_name} OR "Claude Code"
   ```

2. **Fetch Relevant Articles**
   ```
   WebFetch: https://www.claude.com/blog/{article-slug}
   Prompt: Extract practical guidance, examples, and best practices for {topic}
   ```

3. **Cross-Reference with Documentation**
   - Official docs: https://docs.claude.com/en/docs/claude-code/
   - Compare blog insights with documentation

### Monthly Article Review Process

For staying current with Claude improvements:

1. **Fetch Blog Index**
   ```
   WebFetch: https://www.claude.com/blog/
   Prompt: List all articles from the past month with titles, dates, and relevance to Claude Code
   ```

2. **Triage by Relevance**
   - High: Directly mentions Claude Code, CLAUDE.md, Skills, or Agents
   - Medium: General AI coding patterns, prompt engineering
   - Low: Unrelated topics (enterprise, safety research, etc.)

3. **Extract Key Insights**
   For high-relevance articles, extract:
   - New features or capabilities
   - Updated best practices
   - Example patterns and workflows
   - Configuration recommendations

4. **Update Local Knowledge**
   - Consider updating relevant Skills with new patterns
   - Update CLAUDE.md files with new best practices
   - Create new Skills for newly documented features

## CLAUDE.md Best Practices (from Blog + Official Docs)

Based on official documentation and blog posts:

### Structure Guidelines

- **Keep concise**: Treat as documentation both humans and Claude need to understand quickly
- **Strategic placement**: Root for team-wide, parent dirs for monorepos, home folder for universal
- **Use `@import`**: Reference existing docs with `@path/to/file.md` instead of duplicating content
- **Use `CLAUDE.local.md`**: Personal preferences go here (auto-gitignored), not in shared CLAUDE.md
- **Use `.claude/rules/`**: Path-specific rules with `paths` frontmatter for conditional loading

### Memory Hierarchy

| Layer | Purpose | Managed by |
|-------|---------|------------|
| CLAUDE.md | Team instructions (checked into git) | Developers |
| CLAUDE.local.md | Personal preferences (gitignored) | Individual |
| .claude/rules/ | Path-specific and domain rules | Developers |
| ~/.claude/rules/ | User-level rules (all projects) | Individual |
| Auto memory | Claude's own learnings | Claude (automatic) |

### Essential Sections (team-shared)

1. **Project Overview**: Brief summary, key technologies, architectural patterns
2. **Directory Map**: Visual hierarchy showing key directories
3. **Standards & Conventions**: Type hints, code style, naming conventions
4. **Common Commands**: Frequently-used commands with descriptions
5. **Workflows**: Standard approaches for different task types
6. **Tool Integration**: MCP servers, custom tools with usage examples

### What Auto Memory Handles (omit from CLAUDE.md)

- Current focus and work-in-progress context
- Key file relationships and navigation patterns
- Debugging insights and workarounds
- Personal preferences and habits

### What to Avoid

- Sensitive information (API keys, credentials, connection strings)
- Excessive length (keep concise from context engineering perspective)
- Theoretical content not matching actual development reality
- Content that duplicates auto memory concerns

### Evolution Strategy

- Start simple, expand deliberately
- Add sections based on real friction points
- Maintain as living document that evolves with codebase
- Use `@import` to keep CLAUDE.md lean as the project grows

## Delegation Pattern

When research requires deep investigation:

```markdown
Use research-documentation agent for:
- Comprehensive blog article analysis
- Cross-referencing multiple sources
- Building knowledge summaries

Example delegation prompt:
"Research the Claude blog for articles about {topic} from the past 3 months.
Extract practical patterns and update recommendations. Focus on:
- New features or capabilities
- Changed best practices
- Concrete examples we can apply"
```

## Recent Articles Checklist

*Last reviewed: February 2026. Update this list monthly when reviewing new articles.*

Official documentation worth checking (authoritative reference):

- [ ] /docs/en/memory - Memory management, CLAUDE.md, @imports, auto memory, rules hierarchy
- [ ] /docs/en/configuration - Settings, permissions, MCP servers
- [ ] /docs/en/hooks - Hook system for tool use events

Articles worth checking (sorted by relevance to Claude Code):

- [ ] /blog/using-claude-md-files - CLAUDE.md best practices
- [ ] /blog/how-anthropic-teams-use-claude-code - Internal usage patterns
- [ ] /blog/improving-frontend-design-through-skills - Skills feature
- [ ] /blog/claude-code-on-the-web - Web-based features

## Integration with Other Skills

This skill complements:
- **project-discovery**: Use blog patterns for new codebase orientation
- **blueprint-development**: Apply latest CLAUDE.md best practices to PRDs
- **multi-agent-workflows**: Incorporate blog-recommended delegation patterns