This entry is based on the wp-performance skill from the official WordPress/agent-skills repository. The tool behavior is strongly skill-shaped: the agent measures a reproducible target, inspects available profiling tools, and then works through a bounded troubleshooting flow for backend performance issues. It prioritizes WP-CLI, logs, headers, and profiling data over UI guessing, which makes it well suited for coding agents and ops-oriented sessions.

Invoke this when a user reports that a WordPress page, admin screen, REST route, or cron path is slow and wants an agent to diagnose the bottleneck safely. It is specifically useful when the user needs a profiling plan, a backend-only report, and a fix path organized by likely root cause, such as database queries, autoloaded options, object cache misses, remote HTTP calls, or due-now cron spikes.

The scope boundary keeps it from collapsing into a plain product listing. This is not “WP-CLI” as a generic command-line tool, and it is not WordPress as a platform card. The value is the operator playbook: baseline first, inspect capabilities, choose one dominant bottleneck category, verify the delta, and avoid risky production changes without approval. Integration points include WP-CLI profile and doctor commands, Query Monitor headers, curl-based timing, PHP logs, and downstream code changes once the slow path is isolated.