This skill grants broad filesystem and command execution, enabling arbitrary
Claims to do
WP Interactivity API: Use this skill when the user mentions:
Actually does
The skill searches the repository for WordPress Interactivity API patterns (`data-wp-*`, `@wordpress/interactivity`, `viewScriptModule`). It guides on modifying `block.json` or PHP files to enable server-side rendering and initialize state using `wp_interactivity_state()` and `wp_interactivity_data_wp_context()`. It also provides debugging steps, suggests using `@wordpress/create-block-interactive-template`, and refers to internal reference markdown files.
npx skills add https://github.com/wordpress/agent-skillsThe agent is explicitly granted filesystem access and the ability to execute `bash` and `node` commands. This allows for arbitrary code execution, system manipulation, and potential compromise of the host environment.
compatibility: "Filesystem-based agent with bash + node."
The agent has access to `WP-CLI`, a powerful command-line interface for WordPress. This enables administrative actions such as managing users, plugins, themes, and the database, posing a significant risk for site compromise or data manipulation.
compatibility: "Some workflows require WP-CLI."
The skill involves generating and modifying PHP code (e.g., for `wp_interactivity_state` and `wp_interactivity_data_wp_context`). Combined with filesystem access, this capability could be abused to inject malicious PHP code into WordPress files, leading to backdoors or persistent compromise.
wp_interactivity_state( 'myPlugin', array( ... ));
The agent is expected to verify and potentially interact with build tools (`@wordpress/scripts`, webpack/vite). This involves executing `node` or `bash` commands, which could be exploited for resource abuse (e.g., infinite loops, excessive compilation) or to inject malicious build steps.
Verify the repo supports the required module build path: if it uses `@wordpress/scripts`, prefer its conventions. if it uses custom bundling, confirm module output is supported.
The agent is instructed to add or extend Playwright E2E tests. If the agent can control Playwright scripts, this could be leveraged for reconnaissance (e.g., browsing internal web applications, collecting sensitive information) or even data exfiltration through automated browser actions.
If tests exist: add/extend Playwright E2E around the interaction path.
[](https://mondoo.com/ai-agent-security/skills/github/wordpress/agent-skills/wp-interactivity-api)<a href="https://mondoo.com/ai-agent-security/skills/github/wordpress/agent-skills/wp-interactivity-api"><img src="https://mondoo.com/ai-agent-security/api/badge/github/wordpress/agent-skills/wp-interactivity-api.svg" alt="Mondoo Skill Check" /></a>https://mondoo.com/ai-agent-security/api/badge/github/wordpress/agent-skills/wp-interactivity-api.svgSkills can read files, run commands, and access credentials. Mondoo helps organizations manage the security risks of AI agent skills across their entire fleet.