Damage Control Skill

Defense-in-depth protection system for Claude Code. Blocks dangerous commands and protects sensitive files via PreToolUse hooks.

Overview

This skill helps users deploy and manage the Damage Control security system, which provides:

• Command Pattern Blocking: Blocks dangerous bash commands (rm -rf, git reset --hard, etc.)
• Ask Patterns: Triggers confirmation dialog for risky-but-valid operations (ask: true)
• Path Protection Levels:
  • zeroAccessPaths - No access at all (secrets/credentials)
  • readOnlyPaths - Read allowed, modifications blocked
  • noDeletePaths - All operations except delete

Skill Structure

.claude/skills/damage-control/
├── SKILL.md                     # This file
├── patterns.yaml                # Shared security patterns (single source of truth)
├── cookbook/
│   ├── install_damage_control_ag_workflow.md
│   ├── modify_damage_control_ag_workflow.md
│   ├── manual_control_damage_control_ag_workflow.md
│   ├── list_damage_controls.md
│   ├── test_damage_control.md
│   └── build_for_windows.md
├── hooks/
│   ├── damage-control-python/   # Python/UV implementation
│   │   ├── bash-tool-damage-control.py
│   │   ├── edit-tool-damage-control.py
│   │   ├── write-tool-damage-control.py
│   │   ├── python-settings.json
│   │   └── test-damage-control.py
│   └── damage-control-typescript/  # Bun/TypeScript implementation
│       ├── bash-tool-damage-control.ts
│       ├── edit-tool-damage-control.ts
│       ├── write-tool-damage-control.ts
│       ├── typescript-settings.json
│       └── test-damage-control.ts
└── test-prompts/                # Test prompts for validation
    ├── sentient_v1.md
    ├── sentient_v2.md
    ├── sentient_v3.md
    └── sentient_v4.md

After Installation

The install workflow copies hooks and creates settings based on the chosen level:

Global Hooks

~/.claude/
├── settings.json                      # Hook configuration
└── hooks/
    └── damage-control/
        ├── patterns.yaml
        ├── bash-tool-damage-control.py (or .ts)
        ├── edit-tool-damage-control.py
        └── write-tool-damage-control.py

Project Hooks

<agents current working directory>/
└── .claude/
    ├── settings.json                  # Hook configuration (shared)
    └── hooks/
        └── damage-control/
            ├── patterns.yaml
            ├── bash-tool-damage-control.py (or .ts)
            ├── edit-tool-damage-control.py
            └── write-tool-damage-control.py

Project Personal Hooks

<agents current working directory>/
└── .claude/
    ├── settings.local.json            # Personal overrides (gitignored)
    └── hooks/
        └── damage-control/
            ├── patterns.yaml
            ├── bash-tool-damage-control.py (or .ts)
            ├── edit-tool-damage-control.py
            └── write-tool-damage-control.py

---

Cookbook

This section defines the decision tree for handling user requests. Based on what the user says, read and execute the appropriate workflow prompt.

Installation Pathway

Trigger phrases: "install damage control", "setup security hooks", "deploy damage control", "add protection"

Workflow: Read and execute cookbook/install_damage_control_ag_workflow.md

Modification Pathway

Trigger phrases: "help me modify damage control", "update protection", "change blocked paths", "add restricted directory"

Workflow: Read and execute cookbook/modify_damage_control_ag_workflow.md

Manual Control Pathway

Trigger phrases: "how do I manually update", "explain damage control config", "show me the settings"

Workflow: Read and execute cookbook/manual_control_damage_control_ag_workflow.md

Testing Pathway

Trigger phrases: - "test damage control", - "run damage control tests", - "verify hooks are working" - "damage control test this command " - "damage control test this read to this path " - "damage control test this write to this path " - "damage control test this delete to this path " - "damage control test this run this command "

Workflow: Read and execute cookbook/test_damage_control.md

What it does:

• Reads patterns.yaml to get all configured patterns and paths
• Tests PreToolUse hooks (bash, edit, write) with exit code validation
• Tests ask patterns with JSON output validation
• Reports pass/fail for each test case
• Provides summary of all results

Windows Build Pathway

Trigger phrases: "build for windows", "add windows patterns", "convert to windows", "windows damage control"

Workflow: Read and execute cookbook/build_for_windows.md

What it does:

• Checks for existing installation
• Adds Windows PowerShell and cmd patterns alongside Unix patterns
• Creates cross-platform patterns.yaml that works on both systems

Direct Command Pathway

Trigger phrases: "update global read only paths to include X", "add /secret to zero access paths", "block command Y"

Action: Execute immediately without prompts - the user knows the system.

Examples:

• "add ~/.credentials to zero access paths" → Edit patterns.yaml directly
• "block the command 'npm publish'" → Add pattern to bashToolPatterns
• "make /var/log read only" → Add to readOnlyPaths

---

Quick Reference

Settings File Locations

Level	Path	Scope
Global	~/.claude/settings.json	All projects
Project	.claude/settings.json	Current project (shared)
Project Personal	.claude/settings.local.json	Current project (personal)

Path Protection Levels

Type	Read	Write	Edit	Delete	Use Case
zeroAccessPaths	No	No	No	No	Secrets, credentials
readOnlyPaths	Yes	No	No	No	System configs, history
noDeletePaths	Yes	Yes	Yes	No	Important project files

Runtime Requirements

Implementation	Runtime	Install Command
Python	UV (Astral)	curl -LsSf https://astral.sh/uv/install.sh | sh
TypeScript	Bun	curl -fsSL https://bun.sh/install | bash && bun add yaml

Exit Codes

Code	Meaning
0	Allow operation
0	Ask (JSON output triggers dialog)
2	Block operation

---

Testing

Use the test prompts in test-prompts/ to validate the hooks:

• sentient_v1.md - Tests rm -rf blocking (bashToolPatterns)
• sentient_v2.md - Tests find -delete blocking (noDeletePaths)
• sentient_v3.md - Tests ask patterns (SQL DELETE with ID)
• sentient_v4.md - Tests simple command blocking

Run a test:

/project:test-prompts/sentient_v1

---

Related Files

• cookbook/install_damage_control_ag_workflow.md - Installation workflow
• cookbook/modify_damage_control_ag_workflow.md - Modification workflow
• cookbook/manual_control_damage_control_ag_workflow.md - Manual guidance
• cookbook/list_damage_controls.md - List all configurations
• cookbook/test_damage_control.md - Test all hooks
• cookbook/build_for_windows.md - Add Windows patterns
• hooks/damage-control-python/ - Python implementation
• hooks/damage-control-typescript/ - TypeScript implementation