The Agent Skills Directory

COMMAND_EXECUTION (LOW): The skill utilizes high-privilege commands including wp profile eval for dynamic PHP execution and wp db query for direct SQL operations. These are necessary for deep performance profiling but represent a large attack surface. It also references a local script perf_inspect.mjs which was not provided in the skill files.
EXTERNAL_DOWNLOADS (LOW): The instructions suggest installing WP-CLI packages (wp-cli/doctor-command and wp-cli/profile-command) from GitHub. As the wp-cli organization is not on the trusted whitelist, these are categorized as unverifiable external dependencies.
PROMPT_INJECTION (LOW): The skill exhibits an indirect prompt injection surface. 1. Ingestion points: Data is read from external WordPress REST API headers and JSON envelopes (references/query-monitor-headless.md). 2. Boundary markers: Absent. 3. Capability inventory: The agent can execute arbitrary PHP, SQL, and install packages via WP-CLI. 4. Sanitization: No explicit validation or sanitization of ingested REST data is defined.

wp-performance