The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The setup script attempts to acquire administrative privileges to install system dependencies. Evidence: scripts/setup.sh contains sudo apt-get update and sudo apt-get install -y nodejs npm commands.
[DYNAMIC_EXECUTION]: The main script dynamically loads and executes JavaScript modules from a path determined at runtime, which can be manipulated via environment variables or profile settings. Evidence: scripts/wp-content.mjs uses await import() to load a module from a location derived from WP_CLI_PATH or a profile's cli_path property.
[COMMAND_EXECUTION]: The skill performs shell command execution during setup in directories that can be externally defined. Evidence: scripts/setup.sh and scripts/setup.ps1 execute npm install and npm run validate within a directory path obtained from environment variables or configuration files.
[INDIRECT_PROMPT_INJECTION]: The skill fetches and displays content from an external WordPress API, creating a vulnerability where malicious instructions in posts could influence the agent's behavior. Ingestion points: client.listPosts and client.getPost in scripts/wp-content.mjs ingest post data from the WordPress REST API. Boundary markers: The skill does not implement delimiters or instructions to ignore embedded commands within the fetched WordPress content. Capability inventory: The skill environment has capabilities for shell command execution (npm, node), file system access, and network operations. Sanitization: There is no evidence of HTML sanitization, markdown escaping, or instruction filtering for the content retrieved from the WordPress API.

wordpress-content-manager