The Agent Skills Directory

[SAFE]: The skill functions exclusively as a reference guide and template repository for plugin development. It provides accurate architectural guidance without introducing executable risks.
[COMMAND_EXECUTION]: The examples within the skill (such as examples/standard-plugin.md and examples/advanced-plugin.md) reference the execution of local scripts for linting, testing, and security scanning (e.g., bash ${CLAUDE_PLUGIN_ROOT}/hooks/scripts/security/scan-secrets.sh). These operations are constrained to the plugin's own directory using the platform-provided ${CLAUDE_PLUGIN_ROOT} variable and represent standard developer workflows.
[CREDENTIALS_UNSAFE]: The skill follows security best practices by demonstrating the use of environment variable placeholders (e.g., ${API_KEY}, ${GITHUB_TOKEN}, ${KUBECONFIG}) for managing sensitive configuration in MCP server definitions, avoiding hardcoded secrets.

Plugin Structure