Custom Skills & Hooks

Skills: Reusable Prompts

Skills are reusable prompt templates stored in ~/.claude/skills/ or project-level skills/ directories. Instead of retyping complex instructions, you invoke a skill with /skill-name.

Creating a Skill

Create ~/.claude/skills/code-review.md:

# Code Review

Review the specified code for:
1. Logic errors and edge cases
2. Security vulnerabilities
3. Performance issues
4. Type safety
5. Code style consistency

Output a ranked list of findings with severity (Critical/High/Medium/Low)
and suggested fixes for each.

Invoke it with: /code-review src/lib/auth.ts

Hooks: Lifecycle Automation

Hooks run automatically at defined points in the Claude Code lifecycle. Configure them in ~/.claude/settings.json:

{
  "hooks": {
    "PostToolUse": [
      {
        "matcher": "Edit",
        "hooks": [
          {
            "type": "command",
            "command": "npx prettier --write $CLAUDE_FILE_PATH"
          }
        ]
      }
    ]
  }
}

This automatically formats every file after Claude edits it.

Hook Types

| Hook | Trigger | Use Cases | |------|---------|-----------| | PreToolUse | Before any tool runs | Validation, parameter injection | | PostToolUse | After any tool runs | Auto-format, type-check, tests | | Stop | Session ends | Final verification, cost logging | | PreCompact | Before context compaction | Save session state |

Practical Hook Examples

Auto-typecheck after edits:

{
  "matcher": "Edit(*.ts,*.tsx)",
  "hooks": [{"type": "command", "command": "npx tsc --noEmit 2>&1 | tail -5"}]
}

Block dangerous operations:

{
  "matcher": "Bash",
  "hooks": [{"type": "block", "condition": "command contains 'rm -rf'"}]
}

Building a Skill Library

Start a skills/ directory in your project with skills for your most repeated tasks:

/deploy-check — pre-deployment verification checklist
/migration — database migration workflow
/component — React component creation with tests
/api-endpoint — REST endpoint scaffold