claude-hooks-template 1.4.1

CLI tool for configuring Claude Code hooks in projects

claude-hooks

Simple CLI tool for configuring Claude Code hooks in your projects.

What This Does

claude-hooks helps you quickly add hooks to your Claude Code project. Instead of manually:

• Creating .sh hook scripts
• Making them executable
• Editing .claude/settings.local.json

Just run: claude-hooks add <hook-name>

Installation

pip install claude-hooks

Or with uv:

uv pip install claude-hooks

Usage

Navigate to your Claude Code project directory (must have .claude/ directory) and run:

# Install all hooks at once
claude-hooks install

# Or add hooks one at a time:

# Add hook for Edit tool
claude-hooks add on_edit

# Add hook for TodoWrite tool
claude-hooks add on_todo_complete

# Add hook for Stop event
claude-hooks add on_stop

# Add hook for SessionStart event
claude-hooks add on_start

# Add hook for Bash tool
claude-hooks add on_bash

# Add hook for git commits (filter in script)
claude-hooks add on_git_commit

# Add hook for AskUserQuestion tool
claude-hooks add on_ask_user

# Add hook for Write tool
claude-hooks add on_write

# Add hook for MultiEdit tool
claude-hooks add on_multiedit

# Add hook for Task tool (subagent completion)
claude-hooks add on_task_complete

# Add hook for Read tool
claude-hooks add on_read

# Add hook for user prompt submission
claude-hooks add on_user_prompt

# Add hook for plan mode exit
claude-hooks add on_plan

# Add hook for pre-compaction (runs before context compaction)
claude-hooks add on_precompact

# Add hook for periodic snapshots
claude-hooks add on_snapshot

What It Does

When you run claude-hooks add <hook-name>:

1. Creates hook script: .claude/hooks/<hook-name>.sh with basic template
2. Makes it executable: chmod +x automatically applied
3. Updates settings: Adds hook configuration to .claude/settings.local.json
4. Shows summary: Displays hook event and matcher info

Example

$ claude-hooks add on_edit
Created hook script: .claude/hooks/on_edit.sh
Added hook to settings.local.json
  Event: PostToolUse
  Matcher: Edit

Edit .claude/hooks/on_edit.sh to customize hook behavior

Available Hooks

Hook Name	Claude Event	Matcher	Description
on_edit	PostToolUse	Edit	Triggers after Edit tool completes
on_todo_complete	PostToolUse	TodoWrite	Triggers after TodoWrite tool completes
on_stop	Stop	(none)	Triggers when agent stops responding
on_start	SessionStart	(none)	Triggers when session starts
on_bash	PreToolUse	Bash	Triggers before Bash tool executes
on_git_commit	PreToolUse	Bash	Triggers before Bash tool executes (filter for git commit in script)
on_ask_user	PreToolUse	AskUserQuestion	Triggers before AskUserQuestion tool executes
on_write	PostToolUse	Write	Triggers after Write tool completes
on_multiedit	PostToolUse	MultiEdit	Triggers after MultiEdit tool completes
on_task_complete	PostToolUse	Task	Triggers after Task tool completes
on_read	PreToolUse	Read	Triggers before Read tool executes
on_user_prompt	UserPromptSubmit	(none)	Triggers when user submits a prompt
on_plan	PreToolUse	ExitPlanMode	Triggers before ExitPlanMode tool executes
on_precompact	PreCompact	(none)	Triggers before context compaction begins
on_snapshot	PreToolUse	(none)	Triggers periodically to save conversation snapshots

Hook Script Templates

Each generated hook script includes comprehensive documentation:

• Input JSON Schema: All fields hook receives from Claude Code
• Output JSON Schema: Required/optional fields hook must return
• Exit Code Documentation: What each exit code does
• Practical Examples: Real jq parsing examples you can copy-paste

Example: PreToolUse Hook (on_plan, on_read, on_bash)

#!/bin/bash
# on_plan - Triggers before ExitPlanMode tool executes
#
# INPUT JSON SCHEMA:
# {
#   "session_id": "string",
#   "transcript_path": "string",
#   "cwd": "string",
#   "permission_mode": "default|plan|acceptEdits|bypassPermissions",
#   "hook_event_name": "PreToolUse",
#   "tool_name": "string (e.g., 'Read', 'Edit', 'Bash')",
#   "tool_input": {} (tool-specific parameters)
# }
#
# OUTPUT JSON SCHEMA:
# {
#   "hookSpecificOutput": {
#     "hookEventName": "PreToolUse",
#     "permissionDecision": "allow|deny|ask",
#     "permissionDecisionReason": "string (optional)",
#     "updatedInput": {} (optional - modify tool parameters)
#   },
#   "continue": true|false (optional, default: true),
#   "stopReason": "string (optional)",
#   "suppressOutput": true|false (optional, default: false),
#   "systemMessage": "string (optional)"
# }
#
# EXIT CODES:
#   0 - Normal success (allow tool execution)
#   2 - Block tool execution and show stderr to Claude
#   Other - Error condition
#
# EXAMPLES:
#
#   # Allow tool execution:
#   echo '{"hookSpecificOutput": {"hookEventName": "PreToolUse", "permissionDecision": "allow"}}'
#
#   # Block tool execution:
#   echo '{"hookSpecificOutput": {"hookEventName": "PreToolUse", "permissionDecision": "deny", "permissionDecisionReason": "Restricted file"}}'
#
#   # Parse specific fields:
#   TOOL_NAME=$(echo "$INPUT" | jq -r '.tool_name')
#   FILE_PATH=$(echo "$INPUT" | jq -r '.tool_input.file_path // empty')

INPUT=$(cat)

# Default: allow tool execution
echo '{"hookSpecificOutput": {"hookEventName": "PreToolUse", "permissionDecision": "allow"}}'

exit 0

Example: PostToolUse Hook (on_edit, on_write, on_todo_complete)

#!/bin/bash
# on_edit - Triggers after Edit tool completes
#
# INPUT JSON SCHEMA:
# {
#   "session_id": "string",
#   "transcript_path": "string",
#   "cwd": "string",
#   "permission_mode": "default|plan|acceptEdits|bypassPermissions",
#   "hook_event_name": "PostToolUse",
#   "tool_name": "string (e.g., 'Edit', 'Write', 'TodoWrite')",
#   "tool_input": {} (tool-specific parameters that were used),
#   "tool_response": {} (tool-specific response/result)
# }
#
# OUTPUT JSON SCHEMA:
# {
#   "decision": "block|undefined (optional)",
#   "reason": "string (required if decision is 'block')",
#   "hookSpecificOutput": {
#     "hookEventName": "PostToolUse",
#     "additionalContext": "string (optional - added to Claude's context)"
#   },
#   "continue": true|false (optional, default: true),
#   "stopReason": "string (optional)",
#   "suppressOutput": true|false (optional, default: false),
#   "systemMessage": "string (optional)"
# }
#
# EXIT CODES:
#   0 - Normal success
#   2 - Show stderr to Claude for processing
#   Other - Error condition
#
# NOTE: Tool has already executed when this hook runs!
#
# EXAMPLES:
#
#   # Add context about what was changed:
#   FILE_PATH=$(echo "$INPUT" | jq -r '.tool_input.file_path // empty')
#   echo '{"hookSpecificOutput": {"hookEventName": "PostToolUse", "additionalContext": "Modified: '"$FILE_PATH"'"}}'

INPUT=$(cat)

# Default: do nothing
echo '{}'

exit 0

After generating a hook, edit the .sh file to add your custom logic. All the documentation you need is in the generated file!

Settings Format

Hooks are added to .claude/settings.local.json in this format:

{
  "hooks": {
    "PostToolUse": [
      {
        "matcher": "Edit",
        "hooks": [
          {
            "type": "command",
            "command": ".claude/hooks/on_edit.sh"
          }
        ]
      }
    ]
  }
}

Requirements

• Python 3.10+
• Claude Code project with .claude/ directory
• jq (optional, for parsing JSON in hook scripts)

Development

Setup

# Clone repository
git clone https://github.com/stevennevins/claude-hooks.git
cd claude-hooks

# Install dependencies
uv sync --group dev

Running Tests

# All tests
uv run pytest

# Unit tests only
uv run pytest tests/unit/ -v

# With coverage
uv run pytest --cov=claude_hooks

Code Quality

# Format code
uv run ruff format claude_hooks/ tests/

# Lint code
uv run ruff check claude_hooks/ tests/

# Type check
uv run pyright claude_hooks/

Why This Tool?

Grug tired of manual hook setup. Make simple tool that:

• Create hook script fast
• Update settings automatically
• Work every time same way
• No remember JSON structure

Simple better than complex.

License

MIT