Primary question: How do you structure subagents, commands, skills, hooks, and settings for repeatable Claude Code workflows?
RepoDaily adoption score
RepoDaily rates this as 91/100 (strong) for adoption: evidence, installation path, production risk, differentiation, license clarity, and AI/agent fit are scored from the article sources and adoption notes.
5 source(s) across 4 source category/categories, plus a RepoDaily-specific evidence module when available.
5 workflow step(s), 5 next-action step(s), and 1 command/install signal(s) were detected.
Trending momentum is +450 stars, with maintenance/release/issue signals counted when present.
Risk is marked low, with 4 security note(s) and 3 explicit skip condition(s).
3 opportunity lens item(s), 4 alternative(s), and 0 type-specific section(s) support differentiation.
License source or license wording is present.
6 AI/agent-related signal(s) were detected in the article text and metadata.
Project overview
claude-code-best-practice is a curated knowledge repository that maps Claude Code's sprawling feature set into a single navigable guide. Rather than wrapping an API or shipping a runtime library, it pairs best-practice documentation with ready-to-adapt implementation files for every major capability: subagents, slash commands, skills, workflows, hooks, MCP servers, plugins, settings, status lines, memory, and CLI flags.
The repo organizes content through a badge system—green 'Best Practice' badges link to explanatory markdown, while blue 'Implemented' badges jump to working code in the repository. A dedicated 'Hot' table surfaces newer and beta features such as Ultrareview, Ultraplan, Auto Mode, Advisor, Fast Mode, Computer Use, and the Agent SDK, giving readers a quick scan of the cutting edge.
Maintained by Shayan Rais under an MIT license and sponsored by Disrupt.com and ClaudeKit, the project explicitly positions itself as a bridge from ad-hoc 'vibe coding' to disciplined 'agentic engineering.'
Why it is trending now
- Covers the full Claude Code feature surface—subagents, commands, skills, hooks, MCP, plugins, settings, memory, and CLI flags—in one place
- Pairs every concept with both a best-practice explainer and a working implementation file
- Includes a 'Hot' table tracking beta features like Ultrareview, Ultraplan, Auto Mode, Advisor, and Computer Use
- Badge-driven navigation makes it easy to jump between theory and practice
- References community signals including Boris Cherny's Claude Code tips on X
Problem it solves
- Claude Code's capabilities are documented across many separate pages with no unified playbook
- Developers struggle to move from one-off prompt interactions to repeatable multi-agent workflows
- Teams lack patterns for composing subagents, memory rules, hooks, and MCP servers together
- No single resource shows both the recommended approach and a concrete implementation for each feature
How it works
- Open the README concepts table and identify the Claude Code feature you need (e.g., Subagents, Commands, Skills).
- Click a 'Best Practice' badge to read the explanatory markdown for that feature.
- Click an 'Implemented' badge to view a working configuration file in the repository.
- Adapt the implementation files—such as `.claude/agents/<name>.md` or `.claude/commands/<name>.md`—into your own project's `.claude/` directory.
- Consult the 'Hot' table to explore newer beta capabilities and integrate them into your workflow.
Product demo and interface preview
Core: Subagents, Commands & Skills
- Subagents live in `.claude/agents/<name>.md` for delegating specialized tasks
- Commands live in `.claude/commands/<name>.md` for reusable slash-command workflows
- Skills live in `.claude/skills/<name>/SKILL.md` and include references to official Anthropic skills and monorepo strategies
Integration: Hooks, MCP & Plugins
- Hooks in `.claude/hooks/` with a companion best-practices repo for event-driven automation
- MCP Servers configured via `.claude/settings.json` and `.mcp.json` for external tool connections
- Plugin guidance covers marketplaces and distributable packages
Configuration: Settings, Status Line & Memory
- Settings documented across permissions, model config, output styles, sandboxing, keybindings, and auto mode
- Status Line customization via `.claude/settings.json` with a dedicated implementation repo
- Memory spans `CLAUDE.md`, `.claude/rules/`, and global project memory directories
Hot & Beta Features
- Ultrareview (`/code-review ultra`) and Ultraplan (`/ultraplan`) for advanced code analysis
- Auto Mode (`--permission-mode auto`) and Fast Mode (`/fast`) for accelerated workflows
- Advisor (`/advisor`) and Computer Use for deeper agent capabilities
- Agent SDK via npm/pip and orchestration workflow demos for multi-agent composition
Who should pay attention?
Good fit if
- Teams standardizing Claude Code workflows across multiple projects
- Developers building multi-agent orchestration or complex slash-command pipelines
- Engineers setting up hooks, MCP servers, and memory rules for the first time
- Anyone exploring beta features like Ultrareview, Auto Mode, or Advisor
Skip for now if
- Developers not using Claude Code or the broader Anthropic ecosystem
- Teams looking for a runtime library or importable SDK—this is a guide with example configs, not installable code
- Projects that cannot align with MIT-licensed community resources
Risks and cautions
MIT-licensed documentation and configuration examples with no runtime dependencies or executable integration code.
- No production runtime to deploy—content is reference material and example files
- MIT license permits broad commercial and internal use
- Accuracy depends on community maintenance and Claude Code's evolving feature set
- MIT licensed for transparency and permissive reuse
- Hook and settings examples should be reviewed before applying to production environments
- MCP server configs (`.mcp.json`) should be audited for any external endpoints they reference
- Auto Mode and Computer Use grant elevated permissions—evaluate the risk before enabling
Alternatives to compare
| Approach | When to use | Trade-off |
|---|---|---|
Anthropic Official Skills | When you need production-ready skills maintained directly by Anthropic | Free |
Claude Agent SDK Demos | When you want to build agents programmatically via the SDK rather than through Claude Code config files | Free |
Anthropic Prompt Engineering Tutorial | When you need foundational prompt engineering skills before diving into agentic workflows | Free |
Claude Code Official Docs | When you need the canonical reference for any Claude Code feature | Free |
What this trend reveals
Team Onboarding Playbooks
Adapt the best-practice patterns into an internal onboarding guide so new hires can productive with Claude Code within hours instead of days.
Pilot with one squad, measure time-to-first-merged-PR using Claude Code, then roll out.
Domain-Specific Skill Library
Use the skills implementation patterns as a template to build a library of domain-specific SKILL.md files tailored to your codebase and review standards.
Identify the top three repetitive code tasks in your team and create a skill for each.
Multi-Agent Orchestration Prototyping
Leverage the orchestration workflow examples to prototype a multi-agent pipeline for tasks like code review, test generation, or documentation drafting.
Start with the weather-orchestrator command as a structural reference and substitute your own agent roles.
RepoDaily verdict
A well-organized, badge-driven field guide that turns Claude Code's scattered documentation into actionable patterns—valuable for any team serious about moving from experimental vibe coding to disciplined agentic engineering.