Product Tour / Security

Stop unsafe AI requests before they run.

Prompt Firewall reviews AI requests for practical security problems such as secrets, jailbreak attempts, prompt injection, unsafe tools, and off-domain agent behavior.

Start with CloudAll product screens
Prompt Firewall

How this screen helps you understand company AI.

Inline security decisions for jailbreaks, prompt injection, secrets, unsafe tools, and domain abuse.

How the application works

  • SignalCost evaluates eligible chat, tool, and agent invocations before the provider call completes.
  • Rules inspect the request for known security and data-handling signals.
  • The result is recorded with a decision, explanation, and receipt.

Why business owners care

  • Reduces the risk of employees or agents sending unsafe instructions.
  • Gives security teams evidence for why a request was blocked.
  • Protects public-facing agent workflows from off-domain tool abuse.
What You Can Do

A practical workflow, not a static screenshot.

These are the actions customers use inside SignalCost after signup.

Review security decisions.

This action is available from the Security area of the SignalCost workspace.

Tune allowed tools, domains, and rules.

This action is available from the Security area of the SignalCost workspace.

Open evidence for blocked or risky activity.

This action is available from the Security area of the SignalCost workspace.

Related Screens
Console OverviewClean start dashboardCapture CoverageProvider coverageBeacon FleetBeacon FleetTraffic WorkbenchLive traffic by business area
Start with SignalCost

See what AI spend is for before it becomes another bill.

Start with Cloud. Add Beacon One when an office needs local control.

Create workspaceExplore Beacon