AgentGuard

Explainable browser automation signals, local audit and sensitive-domain protection.

Understand browser automation signals and take control of affected tabs. AgentGuard helps you recognize when browser automation may be operating in your Chrome tabs. It presents supporting evidence, heuristic evidence scores and detection coverage so you can understand what was observed—and what remains unknown. KEY FEATURES • Automatic monitoring of accessible websites, even when the popup is closed. • Per-tab assessments with explainable evidence and coverage warnings. • Color-coded toolbar indicators and on-page alerts, including when the extension is unpinned. • Local observation sessions, activity reports, timelines and exportable audit records. • Sensitive-domain policies to monitor, warn or quarantine matching tabs. • Emergency controls to quarantine or close detected tabs. • Optional network blocking and traffic-metadata observation, subject to Chrome permissions and browser restrictions. • Managed configuration support for organizations using the same extension. LOCAL PROCESSING AND DATA HANDLING Core detection runs locally. AgentGuard does not record page contents, passwords, cookies, authentication tokens or form values. Local records can include hostnames, tab identifiers, timestamps, evidence and bounded activity counts. Optional external services are disabled by default and require explicit configuration and user action. IMPORTANT LIMITATIONS Automation signals do not prove that an AI agent controls a tab or identify a specific AI product. Developer tools and legitimate automation can produce similar indicators. Restricted pages, withheld site access and organization policies may reduce detection coverage. A green icon or “No automation observed” is not proof of human control or safety. Quarantine affects the browser tab; it does not terminate an external controller or undo earlier actions.