Overview
Sensitive data → destination hosts
Each ribbon connects a kind of data to the host it was sent to. Width is the number of times; the solid part was stopped, masked or redirected. Hover for details, click a host to allow it.
Risk, day by day
Daily risk indicator, 0 to 100, from the weight of the events of each day.
Users with the most weight
Weight is the sum of the risk points of a user's events in the period.
Logs
The content people typed or uploaded is never stored: only the kind of data, the site and what happened.
| Time | Host | Browser | Policy | Detected topics | Action | Outcome | Site category | User | Device | Note |
|---|
Risky profiles
Weekly risk score
Incidents by risk
Risk score over time
Users by risk
Incidents
| Event name | Incidents | Risk level |
|---|
Riskiest users
| User | Incidents | Risk level | Weight |
|---|
| Timestamp | Source host | Outcome | User | Device | Note |
|---|
Threats
Phishing pages stopped on the page the user actually saw, look-alike addresses corrected, and sites blocked by the web filter. Reported by the browsers themselves.
Extension vetting
Extensions by warning level
Devices
Policy
Extensions marked "always allow" or "ignore" are never disabled.
| Name | Warning level | Threat flags | What it can do | Devices | Users | Browsers | Status | Actions |
|---|
Policy
Browsers pick up a saved policy within five minutes. Every field below is written to the same policy file the browsers read.
General
Classification service
With a service address, prompts and files are classified by the company's own classification service (a multilingual model on the company's server) instead of the keyword lists in the browser. The text is processed in memory and never stored. Leave empty to classify in the browser only.
Teach the service with the company's own documents
Paste examples of one kind of data, or pick text files: each example becomes a reference the service compares new text with. Twenty to fifty examples per kind make a difference. The examples are kept on the classification service only, never in this console. Save the policy with the admin key first.
AI tools on the computer
Applied by the agent on managed computers within five minutes. The gateway choice needs the provider accounts and the people's keys from the Setup tab.
Approved tools people are sent to
Contextual DLP: nine kinds of company data on AI tools
What happens when a prompt, a paste or an uploaded file about one of these topics goes to an AI tool that is not approved.
| Kind of data | Example | What happens | Message shown |
|---|
Detections, whatever the topic
| Found in the text | What happens | Message shown |
|---|
Chat apps that are not approved
Web data loss prevention
Controls on ordinary websites. Leave the domains empty to apply everywhere except the company's work tools; list domains to apply only there.
Watermark
Account protection
Threat protection
Protected brands
Brands the phishing and look-alike checks know. Add the company itself, customers, suppliers and partners.
Advanced: the whole policy as JSON
Lists
Whitelist
Hosts that get no checks at all: no prompt analysis, no phishing check, no watermark. One per line.
Blacklist
Hosts the browser must not open. The user sees the stop page. One per line.
Work tools
Company tools: uploads, downloads and the company account are allowed there, and they are never treated as phishing. One per line.
Devices
Managed computers
Computers that run the agent. The agent keeps the browser protection installed in every browser and reports the state of the computer. Install it once per computer; nothing has to be typed.
| Computer | System | Status | Management | Browser protection | Antivirus | Firewall | Disk encryption | AI tools found | Last seen |
|---|
Add a computer
Download the installer for this company and run it on the computer as administrator. The computer appears here within a few minutes and the browser protection follows by itself.
Connection to the device management service (operator)
Browsers
| User | Device | Browser | Enrolled | Last seen | Status |
|---|
Admin actions
Tenants
One tenant per company. Each has its own policy, devices, events, admin key and enrollment key; a company's admin sees only its own tenant. You see them all.
Companies
| Name | Slug | Tier | Risk, 7 days | Devices | Missing | Events, 30 days | Stopped | Admins | Last event | Actions |
|---|
New tenant
Keys of the new tenant
The admin key is shown only now. Give it to the company's administrator and keep it somewhere safe.
Operator actions
Operators
Accounts that see every company, like the operator key. Give them one-time codes.
| Name | One-time codes | Last sign-in | Actions |
|---|
Setup
Connect a browser
- In Chrome or Edge open
chrome://extensions, switch on Developer mode, choose Load unpacked and pick the project'sextensionfolder. - Open the extension's Settings.
- Enter the work email, a device name, this server's address (
) and the enrollment key below.
Forced install on managed PCs
The Browser Guard installer writes the browser policies that install the extension from this server and keep it installed: no remove button, no developer mode, no incognito or guest window. Run it as administrator on each PC, or push it with the tool you already use.
node scripts/pack-extension.mjs and restart the server.Scripts: Install-BrowserGuard.ps1. The same script with -Uninstall removes everything it wrote. Chrome and Edge accept a forced install from this server only on managed PCs (domain-joined, Entra-joined, MDM-enrolled or Chrome Browser Cloud Management); on other PCs the extension must come from the stores, which is the installer's default source.
Licence
Enterprise components are delivered separately; the tier switch is recorded in the audit log.
Privacy settings
Switching this off shows email addresses everywhere in the console. The change is written to the audit log.
Alerts
Every five minutes the server sends the critical events (phishing, warnings bypassed, risky extensions) and the devices that stopped reporting. Never the content.
AI gateway
The one door to the AI providers for command-line tools, scripts and apps. Each person gets a company key; the same data rules as in the browser are applied before anything leaves, and the provider's own key never leaves this server. Only the kind of data is recorded, never the text.
Provider accounts of the company
Company keys for people
| Note | Created | Last used |
|---|
Administrators
People who sign in to this console for your company with email, password and, once they switch it on, a code from an authenticator app. A new administrator gets a setup link to choose the password; the link works for seven days and is shown only once.
| Name | One-time codes | Last sign-in | Actions |
|---|
Account
One more step before you can use the console. This server requires a one-time code at every sign-in. Set up your authenticator app below; it takes a minute.
Password
One-time codes
A second check at sign-in: a six-digit code from an authenticator app on your phone (Google Authenticator, Microsoft Authenticator, Aegis, 1Password). Strongly recommended for every administrator.
Off
1. Open the authenticator app and add an account by typing this key:
or open this link on the phone: otpauth link
On Every sign-in asks for a code.