FAQ

Is this built on wp_ai_client_prevent_prompt?

Yes. The hard stop uses WordPress core’s own wp_ai_client_prevent_prompt filter — the sanctioned mechanism the AI Client ships for exactly this purpose. No core patching, no HTTP interception. Returning true from that filter stops a call inside WordPress before any request is built, which is why blocked calls never reach (or get billed by) your provider. Full technical detail: hard stop internals.

What does a plugin see when its call is blocked?

The block happens inside WordPress’s AI pipeline, via the AI Client’s own wp_ai_client_prevent_prompt filter. A blocked plugin receives a standard WP_Error (prompt_prevented) from generate calls, and the AI Client’s capability checks report AI as unavailable — the same signals plugins must already handle for “no provider configured.” Well-built plugins degrade gracefully; even a badly-built one cannot spend, because the request never leaves your site and the provider never sees it. We verified this against a live provider: blocked calls simply don’t appear in the provider’s logs.

Does it cap spend per plugin, or only sitewide?

Both. Each plugin can have its own monthly budget, there’s a default budget for plugins without one, and a sitewide budget on top. They’re enforced independently — one plugin hitting its cap doesn’t affect the others. See the alerts and thresholds reference.

Can I get warnings without blocking anything?

Yes — set budgets and thresholds and leave the hard stop off. Alerts fire either way. This alerts-only week is the rollout we recommend in the budget rollout checklist.

What happens if your license server is down?

Nothing, by design. License checks fail open: on any network error the plugin keeps enforcing your budgets using its cached state (rechecked daily). An outage on our side can never switch your spend protection off.

What happens if I cancel?

Updates and support stop; the free AI Spend Monitor, your dashboard, and your full usage history stay — they’re a separate free plugin and your data lives in your database. Nothing is held hostage.

Why does it require the free Monitor?

The Monitor is the data layer: it records every AI call and attributes it to a plugin. Governance reads that data to know month-to-date spend per plugin. Without it, budgets would have nothing to measure — so activation requires it (WordPress’s standard Requires Plugins mechanism), instead of failing silently.

Are the budget amounts exact?

Token counts are exact (verified against WordPress core’s own AI request logs). Dollar amounts are estimates from published list prices and err high — provider discounts aren’t modeled. If you want enforcement against your contracted rates, set them with the Monitor’s aismon_cost_rates filter; Governance budgets use whatever rates the Monitor uses.

Does it cover plugins that bring their own API key?

No — and neither does anything else at the WordPress layer. Governance controls calls made through WordPress 7.0’s shared AI Client. A plugin calling OpenAI directly with its own key bypasses WordPress entirely. The free Monitor’s 60-second test shows you which of your plugins fall in each camp before you buy.

What does the kill switch do?

Blocks every AI call on the site immediately, regardless of budgets, until you switch it off. Blocked calls are still logged and a suspension notice is emailed. It’s for the moment something is actively going wrong and you want everything paused while you investigate.

How is it licensed and priced?

One license per site; a 5-site bundle is available. $39 for the first year during Founder Early Beta, renewing at the regular $79/yr — renewal price is shown before checkout, with a 30-day full refund. Updates are delivered securely from our update server; the free Monitor updates through WordPress.org once listed.