How FormShield checks contact-form spam

FormShield is a message-classification service for website contact forms. Send a message to its API before emailing it or saving it as a lead. The response helps your application decide whether to allow, review or block the submission.

What problem does it address?

Unsolicited SEO pitches, suspicious promotions and other unwanted messages can arrive even when a visitor passes a bot check. FormShield adds a content signal. It does not verify a sender’s identity or replace your firewall, rate limits or CAPTCHA.

The request and decision

  1. Your server validates the form and sends the message, with optional sender details, to /api/v1/check using a private API key.
  2. FormShield checks the key and available quota, then classifies the content.
  3. Your server receives a verdict, score, reason, engine and remaining quota. Decide what to do before forwarding or storing the message.
How to handle each verdict
VerdictSuggested handling
allowContinue normal submission; this is not a guarantee that the sender is trustworthy.
reviewRoute to your own manual review queue instead of automatically rejecting it.
blockConsider stopping delivery after testing against your own legitimate messages.

FormShield currently provides the decision signal and usage dashboard. It does not provide an inbox or a hosted quarantine queue.

Does every request use AI?

No. The public demo uses local rules. The API supports TypeSafe Jev when that provider is configured and available; otherwise it uses local rules. The response’s engine field identifies the engine used. An API key identifies your account and quota; its name does not prove that a model was called.

What are the integration limits?

Server-side checks keep the API key private and can enforce a decision before accepting a submission. The optional browser script exposes its key and can be bypassed. It lets submissions proceed after a failed request or six-second timeout. The WordPress plugin is a pre-release and needs compatibility testing against your installed versions.

How should I evaluate it?

Start with the free plan’s 100 monthly checks. Use genuine inquiries and unwanted messages from your own site, with unnecessary personal details removed. Review false positives and missed spam separately. We do not publish an accuracy percentage or latency guarantee because a representative independent evaluation has not been completed.

Read the API guidePlan a safe rolloutCompare quotas and pricing