The Agent Skills Directory

[SAFE]: The skill is an educational resource and developer guide for the Claude Agent SDK. It does not contain any malicious instructions or hidden threats.
[COMMAND_EXECUTION]: The skill describes capabilities for running shell commands and file operations within the context of the SDK. It includes a specific security section that demonstrates how to block dangerous commands, such as fork bombs, which is identified as a benign educational example.
[EXTERNAL_DOWNLOADS]: The documentation provides references to official repositories and tools for Anthropic and the Model Context Protocol (MCP). The examples of using npx to load MCP servers are standard for this development environment.
[CREDENTIALS_UNSAFE]: The code snippets follow security best practices by demonstrating the use of environment variables for managing sensitive API keys and database credentials.

claude-sdk