The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external information such as plugin versions and vulnerability data retrieved from WordPress-related tools.
Ingestion points: Untrusted data enters the agent context through wordpress_plugins_list-installed, wordpress_themes_list-installed, and vulnerability reports in wordpress_installations_show-core-version.
Boundary markers: Instructions do not specify explicit delimiters or warnings for embedded instructions in the processed data.
Capability inventory: The skill has capabilities to update core software, manage plugins/themes, and modify server-side cache and maintenance settings.
Sanitization: The skill mitigates the risk of processing untrusted data by requiring the agent to report findings and obtain user approval before executing any write operations.
[COMMAND_EXECUTION]: The skill uses curl to perform HTTP health checks on the user's domain after updates are applied. This is a functional requirement to verify that the website still loads (returning a 200 status code) and has no PHP errors after maintenance tasks.

maintain-wordpress