An eCorp Venture

Your forms are only as safe as the last time someone attacked them.

NetworkAgent is a standing adversary for your own network. It fires real attack payloads at every capture surface you own, on a schedule, and reports which ones still accept them — with evidence, not a green checkmark.

97
surfaces monitored
62
surfaces verified clean
0 exposed
accepting hostile input
2026-09-09 06:30:42
last sweep (UTC)

What it watches

Forms

Every lead, contact, newsletter and quote endpoint. Injection, honeypot, throwaway and undeliverable addresses, impersonation of your own people.

Workers

Edge code that has quietly drifted — a guard removed by a redeploy, a new route with no protection, an endpoint nobody remembered shipping.

Agents

Agent cards and machine interfaces: reachable, schema-valid, and not leaking anything they should not.

Networks

Portfolio-wide coverage. One host per surface, so a change anywhere shows up without flooding anything.

Why a 200 response proves nothing

  1. Attack. A uniquely tagged hostile payload is sent to a real capture endpoint — in more than one encoding, because a guard that only understands JSON is bypassed by posting a form.
  2. Look for the row. A correct guard answers with a benign success on purpose, so the response tells you nothing. The verdict comes from whether the record actually landed in the database.
  3. Clean up. The probe deletes its own markers. Nothing an attack probe sends ever reaches a notification, an auto-reply or a human.
  4. Report. Per host, per sweep, timestamped. Anything that stored an attack is flagged immediately.

Get on the list

NetworkAgent is running across our own network today. Tell us what you run and we will tell you what we find.