Claude Code Cheat Sheet

Skills

Reusable prompt templates that your team can invoke by name or let Claude auto-match by description.

Config: .claude/skills/*.md

Frontmatter

• name — Human-readable skill name
• description — When to use this skill (used for auto-matching)
• allowed_tools — List of tools the skill can use (e.g., Bash, Read, Write)
• user_invocable — true/false — whether users can trigger via /name

---
name: deploy
description: Deploy the current branch to staging or production
allowed_tools:
  - Bash
  - Read
user_invocable: true
---

Deploy the app using the following steps:

1. Run tests: `npm test`
2. Build: `npm run build`
3. Deploy to the environment specified in $ARGUMENTS (default: staging)
4. Verify the deployment health check passes
5. Report the deployed URL back to the user

Custom Agents

Specialized sub-agents that run with their own system prompt and tool restrictions. Ideal for parallel workstreams.

Config: .claude/agents/*.md

Frontmatter

• name — Agent identifier (used in Task tool calls)
• description — What this agent specializes in
• allowed_tools — Whitelist of tools (empty = all tools)
• model — Optional model override (sonnet, opus, haiku)

---
name: test-runner
description: Run tests and report failures with suggested fixes
allowed_tools:
  - Bash
  - Read
  - Grep
  - Glob
model: sonnet
---

You are a test-running specialist. When invoked:

1. Identify the test framework in use (check package.json, pytest.ini, etc.)
2. Run the full test suite
3. For any failures, read the failing test and source code
4. Report: which tests failed, why, and a suggested fix
5. Never modify source code — only report findings

Hooks

Shell commands that fire automatically on lifecycle events. Use them for guardrails, formatting, logging, and notifications.

Config: .claude/settings.json → hooks

Lifecycle Events

Examples

{
  "hooks": {
    "PreToolUse": [
      {
        "matcher": "Bash",
        "command": "bash -c 'read input; echo \"$input\" | jq -r .tool_input.command | grep -qE \"git (push --force|reset --hard|clean -f)\" && exit 1 || exit 0'"
      }
    ]
  }
}

{
  "hooks": {
    "PostToolUse": [
      {
        "matcher": "Write|Edit",
        "command": "bash -c 'read input; FILE=$(echo \"$input\" | jq -r .tool_input.file_path); npx prettier --write \"$FILE\" 2>/dev/null || true'"
      }
    ]
  }
}

{
  "hooks": {
    "Notification": [
      {
        "command": "bash -c 'read input; MSG=$(echo \"$input\" | jq -r .message); osascript -e \"display notification \\\"$MSG\\\" with title \\\"Claude Code\\\"\"'"
      }
    ]
  }
}

MCP Servers

Model Context Protocol servers expose external tools to Claude. Once configured, their tools appear automatically — no prompt engineering needed.

Config: .mcp.json (project root) or ~/.claude/mcp.json (global)

{
  "mcpServers": {
    "postgres": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-postgres"],
      "env": {
        "DATABASE_URL": "postgresql://localhost:5432/myapp"
      }
    },
    "github": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-github"],
      "env": {
        "GITHUB_TOKEN": "${GITHUB_TOKEN}"
      }
    }
  }
}

Popular MCP Servers

Plugins

Plugins are installable packages that bundle skills, agents, hooks, and MCP config into a single unit. Instead of manually wiring each extension mechanism, you install one plugin and everything connects. Think of them as the distribution layer on top of the five mechanisms above.

Config: Managed via /plugin commands

What a plugin contains

A plugin is a Git repository with any combination of skills, agents, hooks, and MCP server config. One install adds all of them to your project. Authors maintain, version, and document them like any open source package.

Marketplaces

Plugins are organized into marketplaces (curated registries of plugins). Add a marketplace, then browse and install plugins from it.

Build vs. install

You can build your own plugins to distribute your team's setup across repos, or install community plugins for common workflows like TDD pipelines, code analysis, or memory systems.

Commands

Why Your Org Should Build Custom Plugins

The ROI of investing in Claude Code customization.