Claude Code Commands: Patterns

What Are Claude Code Commands?

Slash commands give developers a way to extend Claude Code with custom actions. Type / followed by a command name, and Claude executes a predefined workflow. Across the ClaudePluginHub directory, there are currently 58,466 registered commands spanning 31,517 plugins. That makes commands the third-largest component type out of 292,892 total components in the ecosystem.

Commands range from simple code-review triggers to complex multi-step workflows. The commands directory lets you browse and filter them by capability, but understanding the patterns behind high-traction command plugins helps you evaluate what to install—or what to build yourself.

How Slash Commands Work

A slash command is a named entry point that Claude Code recognizes when you type /commandname. Plugins register commands in their configuration, and each command can:

• Accept parameters and context from the current session
• Chain together multiple tool calls (file reads, edits, shell commands)
• Trigger agents or skills defined in the same plugin
• Produce structured output or modify files directly

Commands differ from agents and skills in scope. An agent runs autonomously across multiple steps; a skill provides a reusable capability. A command is an explicit user-triggered action—you invoke it, it runs, it finishes. This makes commands the most predictable extension point for developers who want control over when automation kicks in.

Some plugins ship commands alongside other component types. A plugin with commands, agents, and skills can use a command as the entry point that orchestrates the other components. This layered approach shows up repeatedly in high-traction plugins.

Plugins With Notable Command Libraries

Here are eight plugins from the directory that illustrate different approaches to slash commands. Each is linked so you can inspect its components and install data.

ecc (Everything Claude Code by Affaan)

ecc records 231 installs over the last 7 days and has 186,770 stars. It ships commands alongside agents, skills, hooks, and MCP servers. The plugin provides 300+ components for orchestrating autonomous multi-agent coding workflows—TDD enforcement, security audits, production code generation across JS/TS/Python/Rust/mobile stacks, performance optimization, and deployment automation. Its commands serve as entry points into these larger workflows.

claude-md-management

claude-md-management focuses on auditing and improving CLAUDE.md files. With 67 installs in the last 7 days and 19,657 stars, it ships commands and skills. Its commands let you discover CLAUDE.md files across repositories, evaluate quality against rubrics, generate reports, and apply targeted improvements after approval. It also captures learnings from sessions to propose updates to CLAUDE.md or .claude.local.md files.

claude-hud

claude-hud takes a different approach—its commands are entirely about the developer experience during a session. With 59 installs in the last 7 days and 23,094 stars, it adds a real-time statusline HUD to Claude Code showing context health, tool activity, agent tracking, and todo progress. Commands handle interactive configuration of layout, presets, and displayed elements. This is a commands-only plugin with no agents or skills.

everything-claude-code (burgebj)

everything-claude-code ships commands, agents, skills, hooks, and MCP servers in a scope similar to ecc. It records 58 installs in the last 7 days. Both plugins aim to be comprehensive toolkits, but they come from different authors with different implementation approaches—worth comparing if you are evaluating broad-scope command bundles.

fullstack-dev-skills

fullstack-dev-skills provides 66 specialized skills alongside commands, covering Python, TypeScript, Go, Rust, Java, PHP, React/Next.js, Django/NestJS/Spring Boot, React Native/Flutter, and Kubernetes/Terraform. It records 52 installs in the last 7 days and 9,154 stars. Commands handle Jira epic planning, ticket execution, sprint retrospectives, and Confluence doc synthesis—bridging coding work and project management.

code-review

code-review ships a focused set of commands for automating multi-agent code reviews on GitHub pull requests. With 43 installs in the last 7 days and 19,650 stars, it audits CLAUDE.md files, detects bugs, analyzes git history and prior PRs, reviews code comments, and scores issues by confidence level. This is another commands-only plugin, demonstrating that a single component type can deliver substantial value when the use case is well-scoped.

prompts.chat

prompts.chat is one of the most-starred plugins in the directory at 162,464 stars. It ships commands, agents, skills, and MCP servers for searching, retrieving, improving, and managing thousands of AI prompts and Claude skills from the prompts.chat collection. Commands let you install skills, fill prompt variables, save custom prompts with metadata, and enhance them using AI—making external prompt libraries accessible through slash commands.

ruview

ruview has 59,770 stars and ships commands, agents, and skills. It provides a code review and analysis toolkit where slash commands give you structured feedback on your code and pull requests.

Patterns to Borrow

Looking across these eight plugins, several command design patterns emerge that are worth adopting in your own plugins.

Commands as workflow entry points. Plugins like ecc and everything-claude-code use commands to kick off multi-step processes that involve agents and skills. The command handles user input and context setup; the agent handles execution. If you are building a plugin with multiple component types, this is the standard orchestration pattern.

Single-purpose command plugins. claude-hud and code-review prove that a plugin shipping only commands can still deliver high value. claude-hud focuses entirely on session visibility; code-review focuses entirely on PR analysis. Neither needs agents or skills because their workflows are well-bounded.

Domain-specific command bundles. fullstack-dev-skills and claude-md-management target specific domains (fullstack development and CLAUDE.md management, respectively) and group related commands together. This makes the plugin discoverable as a coherent set rather than a loose collection of unrelated actions.

Prompt and knowledge integration. prompts.chat shows how commands can bridge external knowledge bases into Claude Code. Its commands pull from an external prompt library, making the plugin a gateway to curated content rather than self-contained logic.

Session visibility commands. claude-hud illustrates that commands are not limited to code generation or file manipulation. Any developer-facing interaction during a Claude Code session—real-time dashboards, context health monitoring, configuration of display elements—is a valid command target.

Wrap-up

With 58,466 commands across the directory, the slash command ecosystem is large enough that clear patterns are visible. The eight plugins covered here represent distinct strategies: comprehensive toolkits (ecc, everything-claude-code), single-purpose utilities (claude-hud, code-review), domain-specific bundles (fullstack-dev-skills, claude-md-management), and external integrations (prompts.chat).

If you are evaluating what to install, start by identifying which pattern matches your workflow. Browse the full commands directory to filter by capability, or explore the plugin directory for plugins that combine commands with agents, skills, and other component types.