Wp Performance

WP Performance (backend-only)

When to use

Use this skill when:

• a WordPress site/page/endpoint is slow (frontend TTFB, admin, REST, WP-Cron)
• you need a profiling plan and tooling recommendations (WP-CLI profile/doctor, Query Monitor, Xdebug/XHProf, APMs)
• you’re optimizing DB queries, autoloaded options, object caching, cron tasks, or remote HTTP calls

This skill assumes the agent cannot use a browser UI. Prefer WP-CLI, logs, and HTTP requests.

Inputs required

• Environment and safety: dev/staging/prod, any restrictions (no writes, no plugin installs).
• How to target the install:
  • WP root --path=<path>
  • (multisite/site targeting) --url=<url>
• The performance symptom and scope:
  • which URL/REST route/admin screen
  • when it happens (always vs sporadic; logged-in vs logged-out)

Procedure

0) Guardrails: measure first, avoid risky ops

1. Confirm whether you may run write operations (plugin installs, config changes, cache flush).
2. Pick a reproducible target (URL or REST route) and capture a baseline:
   • TTFB/time with curl if possible
   • WP-CLI profiling if available

Read:

• references/measurement.md

1) Generate a backend-only performance report (deterministic)

Run:

• node skills/wp-performance/scripts/perf_inspect.mjs --path=<path> [--url=<url>]

This detects:

• WP-CLI availability and core version
• whether wp doctor / wp profile are available
• autoloaded options size (if possible)
• object-cache drop-in presence

2) Fast wins: run diagnostics before deep profiling

If you have WP-CLI access, prefer:

• wp doctor check --all (add --spotlight to show only warnings and errors)
• or name the checks for a perf-focused run: wp doctor check autoload-options-size constant-savequeries-falsy constant-wp-debug-falsy

wp doctor check with no check names and no --all exits with an error instead of running anything.

It catches common production foot-guns (autoload bloat, SAVEQUERIES/WP_DEBUG, plugin counts, updates).

Read:

• references/wp-cli-doctor.md

3) Deep profiling (no browser required)

Preferred order:

1. wp profile stage to see where time goes (bootstrap/main_query/template).
2. wp profile hook (optionally with --url=) to find slow hooks/callbacks.
3. wp profile eval for targeted code paths.

Read:

• references/wp-cli-profile.md

4) Query Monitor (backend-only usage)

Query Monitor is normally UI-driven, but it can be used headlessly via REST API response headers and _envelope responses:

• Authenticate (nonce or Application Password).
• Request REST responses and inspect headers (x-qm-*) and/or the qm property when using ?_envelope.

Read:

• references/query-monitor-headless.md

5) Fix by category (choose the dominant bottleneck)

Use the profile output to pick one primary bottleneck category:

• DB queries → reduce query count, fix N+1 patterns, improve indexes, avoid expensive meta queries.
  • references/database.md
• Autoloaded options → identify the biggest autoloaded options and stop autoloading large blobs.
  • references/autoload-options.md
• Object cache misses → introduce caching or fix cache key/group usage; add persistent object cache where appropriate.
  • references/object-cache.md
• Remote HTTP calls → add timeouts, caching, batching; avoid calling remote APIs on every request.
  • references/http-api.md
• Cron → reduce due-now spikes, de-duplicate events, move heavy tasks out of request paths.
  • references/cron.md

6) Verify (repeat the same measurement)

• Re-run the same wp profile / wp doctor / REST request.
• Confirm the performance delta and that behavior is unchanged.
• If the fix is risky, ship behind a feature flag or staged rollout when possible.

WordPress 6.9 performance improvements

Be aware of these 6.9 changes when profiling:

On-demand CSS for classic themes:

• Classic themes now get on-demand CSS loading (previously only block themes had this).
• Reduces CSS payload by 30-65% by only loading styles for blocks actually used on the page.
• If you’re profiling a classic theme, this should already be helping.

Block themes with no render-blocking resources:

• Block themes that don’t define custom stylesheets (like Twenty Twenty-Three/Four) can now load with zero render-blocking CSS.
• Styles come from global styles (theme.json) and separate block styles, all inlined.
• This significantly improves LCP (Largest Contentful Paint).

Inline CSS limit increased:

• The threshold for inlining small stylesheets has been raised, reducing render-blocking resources.

Reference: https://make.wordpress.org/core/2025/11/18/wordpress-6-9-frontend-performance-field-guide/ (opens in a new tab)

Verification

• Baseline vs after numbers are captured (same environment, same URL/route).
• wp doctor check --all is clean (or improved) when applicable.
• No new PHP errors or warnings in logs.
• No cache flush is required for correctness (cache flush should be last resort).

Failure modes / debugging

• “No change” after code changes:
  • you measured a different URL/site (--url mismatch), caches masked results, or opcode cache is stale
• Profiling data is noisy:
  • eliminate background tasks, test with warmed caches, run multiple samples
• SAVEQUERIES/Query Monitor causes overhead:
  • don’t run in production unless explicitly approved

Escalation

• If this is production and you don’t have explicit approval, do not:
  • install plugins, enable SAVEQUERIES, run load tests, or flush caches during traffic
• If you need system-level profiling (APM, PHP profiler extensions), coordinate with ops/hosting.