security-meta

What problem does it solve?

Ensuring the security of AI-assisted development environments, including protection against prompt injection, managing permissions, and configuring sandboxing, is critical but complex. This skill centralizes security guidance, helping users implement robust protections.

Core Features & Use Cases

• Comprehensive Security Guidance: Covers Claude Code's security fundamentals, permission-based architecture, prompt injection protections, and sandboxing (filesystem and network isolation).
• IAM & Permission Management: Guides users through configuring tiered permissions (allow, ask, deny rules), permission modes (default, acceptEdits, bypassPermissions), and tool-specific rules.
• Credential & Enterprise Policy Management: Explains secure credential storage, custom apiKeyHelper scripts, and setting up unoverridable enterprise security policies via managed-settings.json.
• Cloud & IDE Security: Addresses security considerations for cloud execution, VS Code, JetBrains, and devcontainers, along with general security best practices.
• Use Case: A security engineer needs to understand how Claude Code protects against prompt injection and how to configure strict network isolation for Bash commands. This skill provides the keywords to query official-docs for detailed explanations of core protections and sandbox network settings.

Quick Start

Use the security-meta skill to learn how Claude Code protects against prompt injection.