🤖
Subagents
Spawn specialized agents with custom tools, models, and permissions for parallel task execution.
Learn more
Claude CodeBest Practices
Comprehensive guide to Claude Code configuration, workflows, and best practices
Skip to content
At a Glance
52
Doc Pages
60+
Settings
65
Commands
170+
Env Vars
69
Tips
6
Video Talks
Concepts
| Concept | Description | |
|---|---|---|
| Subagents | Spawn specialized agents with custom tools, models, and permissions | |
| Commands | Custom slash commands (.claude/commands/) with YAML frontmatter | |
| Skills | Reusable capabilities (.claude/skills/) with context files | |
| Workflows | Multi-step orchestration: Command → Agent → Skill | |
| Hooks | Event-driven handlers triggered by Claude Code lifecycle events | |
| MCP Servers | Model Context Protocol integrations (Playwright, Context7, DeepWiki) | |
| Settings | 5-level configuration hierarchy from managed to global | |
| Memory | Persistent context via CLAUDE.md with monorepo loading strategies | |
| CLI Flags | 40+ startup flags and environment variables | |
| Power-Ups | Interactive lessons teaching Claude Code features |
Orchestration Pattern
The Command → Agent → Skill pattern is the recommended architecture:
| Component | Role | Example |
|---|---|---|
| Command | Entry point, handles user interaction | /weather-orchestrator |
| Agent | Fetches data with preloaded skills | weather-agent + weather-fetcher |
| Skill | Creates output independently | weather-svg-creator |
Learn more about the orchestration pattern →
Support This Project
If you find these best practices helpful, consider sponsoring to keep the project maintained and growing.
Become a SponsorBuy a Coffee
Monthly Supporter
Gold Sponsor