CNXT AI Cost
Guardrails
CompanyView plans
Free monthly breaker · Pro 1.5 guardrails

Stop runaway AI costs
before they consume the month.

Retry loops, repeated requests, oversized input or output, unexpected expensive models, and runaway Cron jobs can consume tokens in minutes. Pro limits the blast radius by time, request, caller, model, and execution context.

Protect one production hostname — $99/year Start with Free
Monitoring firstRaw prompts never persistedNative WordPress AI Client
GUARDRAILS · ENFORCINGPRO 1.5
CRITICAL

Retry circuit opened

request_retry_circuit
Source
inventory-sync
Context
background
Matching failures
5 / 5
DecisionBlocked before provider call
Illustrative Pro 1.5 protection record
The problem

Runaway spend begins in seconds,
not at month end.

Free is the dependable monthly wall. Pro intervenes earlier when one request, one minute, one caller, one model, or one execution context starts behaving abnormally.

Free: site-wide and per-source monthly USD, request, and token limits.Pro: seven cause-specific guardrails before the final wall.
The controls

Three limits.
One hard stop.

Free enforces site-wide and per-source monthly USD, request, and token limits. Pro adds faster cause-specific controls for bursts, repetition, retries, models, and execution contexts.

01

Estimated USD

Set site-wide or per-source monthly estimated-USD limits in Free using rates you configure. Pro also adds provider/model caps.

Budget$25.00
02

AI call attempts

Count provider attempts—including no-result attempts—so repeated work cannot disappear from the monthly boundary.

1,284/ 2,000 attempts
03

Total tokens

Cap aggregate token usage across supported requests before it compounds into cost.

1.9M/ 3.0M tokens
+

Site-wide and per-source in Free.

Set separate monthly USD, request, and token budgets for the whole site or individual callers.

Entire site $25 Product writer $10 Support assistant $8
Start safely

Monitor first. Enforce when ready.

Observe real usage without blocking requests. When the limits look right, switch to Enforcement and turn the same policy into a guardrail.

Enforcement
Pro 1.5 guardrails

Seven controls for sudden token growth.

Each guardrail is opt-in. Keep the global mode in Monitoring while you check normal traffic, then switch the global mode to Enforcement.

01

Token velocity

Rolling one-minute and one-hour token limits, including conservative reservations for in-flight calls.

02

Source-aware burst control

Apply velocity, retry, and fallback circuits to the caller that is behaving abnormally.

03

Repeated requests

Detect matching requests without persisting raw prompt content.

04

Per-request size

Limit estimated input and requested output tokens, with explicit block or output-clamp behavior.

05

Retry storms

Cool down matching retries and open a source-level fallback circuit after abnormal failures.

06

Model spend

Apply a monthly estimated-USD cap to a resolved provider and model using rates you configure.

07

Execution context

Choose whether guardrails apply to public, signed-in, administration, and background/Cron/CLI requests.

Explore Pro 1.5 →
Pricing

Free protects the month.
Pro contains the incident.

Free sets site-wide and per-source monthly boundaries. Pro adds controls for sudden growth, request patterns, models, contexts, and business-aware response.

FreeFOREVER
$0 / forever

Basic hard-stop protection

When a site-wide or per-source monthly USD, request, or token limit is reached, matching native AI Client requests stop.

  • Site-wide and per-source monthly USD, request, and token enforcement
  • Monitoring and enforcement modes
  • Emergency stop and blocked-call audit
  • Local usage counters and CSV export
Install Free
BEST VALUE
SAVE $21EVERY YEAR
Pro · AnnualANNUAL BILLING
$99 / year
Equivalent to $8.25/month

Pro 1.5 · 7 Guardrails + Smart Protection

Layered controls for sudden token growth, plus probabilistic traffic and revenue signals beneath an absolute hard stop.

  • Everything in Free
  • One-minute, one-hour, and per-request token controls
  • Source-aware burst control and provider/model estimated-USD caps
  • Matching-request and retry-storm circuits
  • Public, signed-in, admin, and background/Cron/CLI scopes
  • Smart signals, alerts, incidents, and blocked-call audit
ONE PRODUCTION HOSTNAMEOne subscription covers one normalized production hostname.
Subscribe — $99/year

$99 is charged now and every year until canceled. No minimum term. Cancel online before the next renewal. Secure checkout by Stripe.

Pro · MonthlyFLEXIBLE
$10 / month

Pro 1.5 · 7 Guardrails + Smart Protection

The same complete Pro feature set with monthly billing.

  • Everything in Free
  • One-minute, one-hour, and per-request token controls
  • Source-aware burst control and provider/model estimated-USD caps
  • Matching-request and retry-storm circuits
  • Public, signed-in, admin, and background/Cron/CLI scopes
  • Smart signals, alerts, incidents, and blocked-call audit
ONE PRODUCTION HOSTNAMEOne subscription covers one normalized production hostname.
Subscribe — $10/month

$10 is charged now and every month until canceled. No minimum term. Cancel online before the next renewal. Secure checkout by Stripe.

By continuing to Stripe, you agree to the Terms and acknowledge automatic renewal. See our Commercial Disclosure.

No setup or shipping fees. Any tax that must be collected separately is shown before order submission. Free, Pro, Agency, and Unlimited are product tier names and remain the same in every language.

Estimated USD uses rates you configure and is not a provider billing cap. Keep provider-side limits as the final backstop.

How it works

Protection in
three deliberate steps.

01

Connect

Keep using the native WordPress AI Client and your configured provider connectors.

02

Set the boundary

In Free, choose site-wide and per-source monthly USD, request, and token limits; configure token rates for estimated-USD enforcement.

03

Let the breaker trip

In Enforcement, Free stops supported requests at any configured monthly boundary. Pro adds faster, cause-specific protections.

Data minimization

Guard the budget.
Not the conversation.

Raw prompts and responses are never persisted. When Pro repetition protection is enabled, selected request fields are read and canonicalized transiently in memory; only a keyed HMAC-SHA256 digest and operational counters are stored locally.

Read the full Privacy Policy

Privacy information includes APPI, GDPR, international transfers, legal bases, retention, and data-subject rights.

Not storedPrompts✓ raw content is never persisted
Not storedResponses✓ stays out of the log
Not storedAPI keys✓ managed by WordPress
Stored locallyUsage counters✓ retention controls included
Built for

For the people who are accountable after launch.

WP

Plugin teams

Add cost boundaries around AI features without building a separate policy layer.

AG

Agencies

Give client sites explicit AI budgets and a visible emergency stop.

OP

Site operators

Understand what AI is consuming and decide how far it is allowed to go.

FAQ

Straight answers.
No hidden fine print.

Which AI requests does it cover?+

Coverage levels: Full coverage on the guarded default WordPress AI Client HTTP transporter adds exact matching-request/retry checks, resolved-model caps, estimated-input checks, and provider-aware block or clamp for supported output fields. Custom or sidecar transports receive Core-only coverage: lifecycle accounting and source-level fallback protection. Calls made directly to provider APIs are outside scope.

Are the cost numbers exact?+

No. They are estimates calculated from configured pricing and recorded usage. Your provider invoice remains authoritative.

Does Monitoring mode block anything?+

No. Monitoring records usage and shows how limits would behave. Switch to Enforcement when you are ready to block requests that cross a limit.

Does it store prompts or AI responses?+

Raw prompts and AI responses are never stored. If Pro repetition protection is enabled, selected request fields are canonicalized only in memory and only a keyed HMAC-SHA256 digest is stored locally.

What versions does it require?+

WordPress 7.0 or later and PHP 7.4 or later.

What does Smart Protection do?+

Smart Protection uses probabilistic Human, Bot, and Unknown classification with optional revenue signals to reduce unnecessary disruption below absolute limits. It cannot guarantee every legitimate request will continue, and configured hard stops always win.

Do Pro plans renew automatically?+

Yes. Monthly Pro renews at $10 each month and Annual Pro at $99 each year until canceled. Annual saves $21 per year.

How do I cancel Pro?+

Use the online cancellation page before your next renewal. There is no fee or required phone call, and access continues through the paid period. Cancel online →

Can one Pro subscription be used on multiple sites?+

One seat covers one normalized production hostname. Different WordPress paths and HTTP/HTTPS on that hostname share the seat. A different hostname—including a www subdomain—requires another seat.

Pro 1.5 · seven guardrails + Smart Protection

Do not give one broken loop
the whole month's budget.

Explore Pro 1.5
Seller & operator

Creator's NEXT INC.

1-9-36-13F Konan, Minato-ku, Tokyo-to 108-0075 Japan
Representative Director: Nozomu Kubota

+81-3-4405-0319
info@cnxt.jp