Hooks Manager

Overview

hooks-manager enables sophisticated workflow automation through Claude Code's hooks system (released June 2025).

Purpose: Create event-driven automation for code formatting, security gates, observability, and validation

Pattern: Task-based (8 operations, one per hook event + management)

Key Innovation: Hooks transform Claude Code from interactive to automated - reducing manual work by up to 92%

Hook Events Available (8):

1. UserPromptSubmit - Before Claude sees prompts
2. PreToolUse - Before tool execution
3. PostToolUse - After tool execution
4. PermissionRequest - When permissions requested
5. Notification - Important events
6. SessionStart - Session initialization
7. SessionEnd - Session cleanup
8. Stop - Conversation termination

---

When to Use

Use hooks-manager when:

• Auto-formatting code (PostToolUse after Write/Edit)
• Security gates (PreToolUse to block dangerous operations)
• Observability integration (send telemetry on all tool calls)
• Validation enforcement (block invalid operations)
• Workflow automation (trigger actions on events)
• Team standardization (enforce coding standards)

Production Impact: Companies report 92% reduction in style review time with PostToolUse formatting hooks

---

Prerequisites

Required

• Claude Code (hooks support added June 2025)
• Basic understanding of event-driven programming

Recommended

• Linter/formatter installed (ESLint, Prettier, Black)
• Security tools (if using security hooks)
• Observability platform (if using telemetry hooks)

---

Hook Operations

Operation 1: Create Formatting Hook (PostToolUse)

Purpose: Auto-format code after AI writes/edits files

Use Case: Enforce coding standards without manual intervention

Process:

1. Choose Formatter:

   # JavaScript/TypeScript
   npm install --save-dev prettier
   
   # Python
   pip install black
   
   # Multi-language
   npm install --save-dev prettier @prettier/plugin-python
   

2. Create Hook Configuration:

   {
     "event": "PostToolUse",
     "tools": ["Write", "Edit"],
     "description": "Auto-format code after Claude writes/edits files",
     "handler": {
       "command": "npx prettier --write \"$FILE\"",
       "timeout": 5000
     },
     "filters": {
       "filePatterns": ["**/*.{ts,tsx,js,jsx,py}"],
       "excludePatterns": ["node_modules/**", "*.min.js"]
     }
   }
   

3. Install Hook:

   # Global hook (all projects)
   mkdir -p ~/.claude/hooks
   echo '[above-json]' > ~/.claude/hooks/auto-format.json
   
   # Project-specific
   mkdir -p .claude/hooks
   echo '[above-json]' > .claude/hooks/auto-format.json
   

4. Test Hook:

   # Have Claude write unformatted code
   # Hook should auto-format after Write/Edit
   # Verify formatted correctly
   

Outputs:

• Hook configuration file
• Auto-formatting on all code changes
• 92% reduction in style review time

Validation:

• Hook file created
• Formatter configured correctly
• File patterns appropriate
• Tested with Claude Code
• Formatting works automatically

Time Estimate: 30-45 minutes

---

Operation 2: Create Security Hook (PreToolUse)

Purpose: Block dangerous operations before execution

Use Case: Prevent accidental production file modifications, destructive commands

Process:

1. Define Security Rules:

   {
     "event": "PreToolUse",
     "tools": ["Write", "Edit", "Bash"],
     "description": "Block dangerous operations",
     "handler": {
       "command": ".claude/hooks/security-check.sh \"$TOOL\" \"$FILE\" \"$COMMAND\""
     },
     "structuredOutput": true
   }
   

2. Create Security Check Script:

   #!/bin/bash
   # .claude/hooks/security-check.sh
   
   TOOL=$1
   FILE=$2
   COMMAND=$3
   
   # Block production file modifications
   if [[ "$FILE" == "/etc/"* ]] || [[ "$FILE" == "/usr/"* ]]; then
     echo '{"continue": false, "stopReason": "Cannot modify system files", "suppressOutput": true}'
     exit 1
   fi
   
   # Block destructive commands
   if [[ "$COMMAND" =~ rm\ -rf\ / ]] || [[ "$COMMAND" =~ dd\ if ]]; then
     echo '{"continue": false, "stopReason": "Destructive command blocked", "suppressOutput": true}'
     exit 1
   fi
   
   # Block curl/wget (security risk)
   if [[ "$COMMAND" =~ curl|wget ]]; then
     echo '{"continue": false, "stopReason": "Network commands disabled", "suppressOutput": false}'
     exit 1
   fi
   
   # Allow operation
   echo '{"continue": true}'
   exit 0
   

3. Install Security Hook:

   chmod +x .claude/hooks/security-check.sh
   # Hook JSON already configured in step 1
   

Outputs:

• Security hook blocking dangerous operations
• Protection against accidental damage
• Audit trail of blocked operations

Validation:

• Hook blocks system file modifications
• Hook blocks destructive commands
• Hoo

...