agentav pilot --days 7 --mode monitor
A 7-day pilot that blocks nothing
See exactly what Agent Antivirus would catch in your agents before it is allowed to stop anything. Your agents keep working exactly as they do today.
- Agent events scanned, agent by agent and day by day
- Every threat found, with masked evidence and its OWASP code
- What enforce mode would have stopped, and which actions would have waited for a person
- Accuracy on every detection your team reviewed
- Verdict times measured on your own traffic
agentav pilot --plan
How the week goes
- Day 0
Connect
We connect to your agents in an afternoon, usually with the monitoring tap, so no handler code changes. Policy starts in monitor mode.
- Days 1 to 6
Watch and review
Detections arrive within seconds. Your team marks each one as a real threat or a false alarm, about five minutes a day.
- Day 7
Report
You get the report: what was seen, what would have been stopped and how accurate it was on everything you reviewed.
- Then
Decide
Switch to enforce, keep watching, or walk away. Turning the tap off is one setting.
agentav pilot --guarantees
Safe to run on production
The pilot is designed so the worst case is that Agent Antivirus sees less, never that your agents do less.
- Monitor-only: verdicts are recorded and nothing is blocked, held or changed.
- Never in the request path: events go to a bounded background queue, and if Agent Antivirus is slow or down they are dropped, not delayed.
- Off with one setting:
AGENTAV_TAP_ENABLED=0, no code change. - Credential fields are redacted before anything leaves your service.
- Status polling and other read-only requests are not recorded.
- Your data is isolated by your own API key, and history is kept for 90 days unless you choose otherwise.
agentav pilot --request
Request a pilot
Tell us what your agents do and what they run on. We reply within one business day to set up a 30-minute call.
Good fits: teams running agents that browse, read email or documents, call tools through MCP, write code, or move money.