Documentation
Everything you need to know about installing, using, and managing LeakSnitch. From first time setup to enterprise administration.
Getting Started
What is LeakSnitch
LeakSnitch is a browser extension that detects and prevents sensitive data from being sent to AI platforms like ChatGPT, Claude, Gemini, and others. It runs entirely in your browser, scanning keystrokes and pasted content for secrets before they ever reach an AI model.
Installing the Extension
The LeakSnitch extension is available on the Chrome Web Store. It supports all Chromium based browsers including Chrome, Edge, Brave, and Opera. Installation takes under 30 seconds and requires no account or configuration to start working. The extension comes bundled with detection patterns, so it begins protecting you immediately after installation.
Creating Your Account
You can create a free account to access the LeakSnitch dashboard. The dashboard provides visibility into detections, security scoring, and analytics across your browser activity. Registration requires only an email address and password. Google single sign on is also available. Enterprise users are invited by their organization administrators.
Dashboard Overview
The dashboard is your central hub for monitoring data leak activity. It displays key performance indicators including blocked data count, security score, AI tool visit statistics, and recent incidents. The dashboard also includes severity breakdowns via donut charts, top policy violations, and interactive charts for daily trend analysis. Enterprise administrators can view organization wide metrics including user risk rankings and department level analytics.
Extension Permissions
LeakSnitch requires specific browser permissions to function. Storage permission allows the extension to cache detection patterns and user preferences locally. Alarms permission enables periodic background checks for pattern updates on the server. Host permissions are limited to the supported AI platform domains where scanning is needed; the extension cannot read data on other websites. These are the minimum permissions required and no data is collected from permissions beyond on screen text on AI platform pages.
Uninstalling the Extension
You can uninstall LeakSnitch at any time from your browser extension settings. In Chrome, navigate to the extensions page, find LeakSnitch, and click Remove. Uninstalling deletes all locally cached patterns and preferences. Your dashboard account and historical detection data remain intact and can be accessed again if you reinstall. There is no lock in; all bundled detection patterns are publicly available and you can reference them even without the extension installed.
How Detection Works
Supported Platforms
LeakSnitch monitors over 18 AI platforms including ChatGPT, Claude, Gemini, Copilot, Perplexity, Grok, DeepSeek, and others. When you visit any of these platforms, the extension automatically activates and begins scanning your inputs in real time. The full list of supported platforms is publicly available.
Detection Methods
The detection engine uses three complementary methods to identify sensitive data. Pattern matching uses curated regular expressions to detect known secret formats like API keys, authentication tokens, private keys, and credentials. Entropy analysis calculates the Shannon entropy of each input to identify high randomness strings that may be obfuscated secrets even when they do not match a known pattern. Contextual analysis examines surrounding words and sentence structure to determine whether detected strings are likely real secrets or false positives.
What Gets Scanned
Every text input on supported platforms is scanned, including form fields, chat text areas, code editors, and contenteditable elements. The extension also scans content pasted from the clipboard, files uploaded through the browser, and content dragged and dropped onto the page. This covers PDFs, Word documents, Excel spreadsheets, and plain text files. Only the most recent portion of each input is analyzed to maintain performance.
When Detection Triggers
Detection happens the moment you attempt to submit content to an AI platform. The extension intercepts the submission event, scans the content, and either allows, blocks, or redacts the data based on the detection result and your configured action preference. This intervention takes under 50 milliseconds and does not slow down normal prompts.
Detection Response Types
Hard Block
Hard block prevents the submission entirely. A modal appears explaining what type of data was detected and its severity level. The user cannot proceed without acknowledging the block. This is the default action for enterprise organizations and is recommended for high sensitivity environments where no data should ever leak.
Redact Mode
Redact mode automatically replaces detected secrets with a placeholder before the content is sent to the AI platform. For example, an API key like sk-proj would be replaced by API KEY REDACTED. This allows users to continue their work without interruption while ensuring secrets are never transmitted. Enterprise administrators can enforce redact mode across their entire organization.
Snooze
The snooze option temporarily disables detection for a five minute window. This is useful when you need to share a piece of data intentionally and understand the risk. After three snoozes on the same platform within a session, the domain is automatically whitelisted for the remainder of the session.
False Positive Reporting
If LeakSnitch incorrectly flags a piece of data as sensitive, you can submit a false positive report with a single click. These reports help improve detection accuracy over time. The false positive report includes the detected text and the platform context, but does not include any surrounding prompt content.
Dashboard Analytics
Security Score
Your security score is calculated based on your detection history, blocked content ratio, and severity distribution. A higher score indicates better protection and fewer data leak incidents. The score is updated in real time and displayed at the top of the dashboard. Enterprise organizations can view the aggregate security score across all members.
Incident Reports
Every detection event is logged and visible in the dashboard incident report. Each report includes the timestamp, the AI platform involved, the type of data detected, the action taken, and the severity level. The actual secret value is never stored. You can search, filter by severity, and export incident reports to CSV or PDF for compliance documentation.
AI Tool Monitoring
The dashboard tracks which AI platforms you visit and how often. This data is displayed in a bar chart showing your top AI tools ranked by visit frequency. It also shows which platforms generated the most detection events, helping you understand where sensitive data is most at risk within your workflow.
Exporting Reports
Incident data can be exported in CSV or PDF formats from the audits page. CSV exports are suitable for spreadsheet analysis and SIEM ingestion. PDF exports are formatted for compliance reporting and audit documentation. Exports include all visible columns and respect any active filters.
Account Deletion and Data Export
You can request full deletion of your account and all associated data by contacting support or through your profile settings. Upon deletion, all personal information, detection metadata, and analytics data are permanently removed from LeakSnitch servers. If you need a copy of your data before deletion, export your incident reports from the audits page. Enterprise users should contact their organization administrator for data deletion requests, as some data is managed at the organization level.
Enterprise Features
Organization Management
Enterprise organizations can manage users through department based grouping. Each organization has a superadmin who can invite members, assign departments, and configure organization wide policies. User activity is aggregated at the department and organization level for comprehensive visibility.
Custom Detection Rules
Enterprise administrators can create custom detection rules tailored to their organization's specific data types. Custom rules support pattern matching for internal secret formats, proprietary identifiers, and compliance specific data types. These rules apply to all members of the organization and are enforced alongside the default detection patterns.
Department Policies
Organizations can define policies per department, specifying which AI tools each department may use and what action should be taken when sensitive data is detected. For example, the engineering department can have stricter policies than marketing, or certain AI platforms can be blocked entirely for specific departments.
Redact Mode Enforcement
Enterprise administrators can enforce redact mode for their entire organization, ensuring that all detected secrets are automatically replaced before any data reaches an AI platform. This provides a balance between security and productivity, as users can continue working without interruptions while sensitive data is automatically protected.
Upgrading to Enterprise
Individual users can upgrade their account to an Enterprise organization by contacting the LeakSnitch sales team. The upgrade process creates an organization that you administer, with the ability to invite team members, create departments, and configure organization wide policies. Existing detection history is preserved during the upgrade. Pricing is based on organization size and required features. Book a demo or contact sales through the website to begin the upgrade process.
Security and Architecture
Zero Trust Data Architecture
LeakSnitch operates entirely within your browser. All detection processing happens locally on your machine. Your sensitive data including proprietary code, personally identifiable information, and authentication secrets never leaves your computer and is never transmitted to LeakSnitch servers. This zero trust architecture eliminates the risk of server side data breaches.
Local Processing
The detection engine runs as a content script injected into each supported AI platform page. It has access only to the text you type or paste into the page. It cannot read your browsing history, bookmarks, passwords, or content on other websites. All pattern matching, entropy calculation, and validation algorithms execute within the browser sandbox.
Pattern Transparency
LeakSnitch publishes all of its detection patterns publicly. Security teams can audit the rules at any time. No black box, no non disclosure agreement required. Enterprise SOC teams can proxy the rules through their own infrastructure for validation. The competitive advantage is in the context aware detection engine, not hidden patterns.
Data Retention
Only metadata about detection events is stored on LeakSnitch servers. This includes the timestamp, user identifier, type of data detected, action taken, and severity level. The actual payload that triggered the detection is immediately discarded on the client side and never persisted. Account information is limited to email address and display name.
Privacy and Terms
LeakSnitch is committed to transparent data handling. The full Privacy Policy and Terms of Service are published on the website and cover data collection practices, user rights, cookie usage, and legal disclaimers. These documents are reviewed and updated regularly to reflect current practices and regulatory requirements. Users are encouraged to review both documents before creating an account or installing the extension.
Changelog
Version 3.0
Released July 2026. Major update with redesigned detection engine, expanded platform support to 18 AI tools, new dashboard with real time KPI cards and severity charts, enterprise features including custom detection rules and department policies, false positive reporting, and improved context aware detection with entropy analysis fallback. The extension manifest was updated to version 3.0.1 for Chrome Web Store compliance.
Previous Versions
Earlier versions established the core detection framework with pattern matching on major AI platforms, basic dashboard analytics, and individual user accounts. Version 2 introduced the free tier and bundled fallback patterns. The extension has been continuously updated with new detection patterns and platform support since its initial release.
Frequently Asked Questions
Does LeakSnitch slow down my browser
No. The detection engine is optimized to run in under 50 milliseconds per scan. Normal prompts are not affected. Detection only triggers on content that matches a sensitive pattern, so the vast majority of your AI interactions proceed at full speed with zero latency.
Can LeakSnitch read all my browser data
No. The extension has permission to run only on supported AI platform domains. It cannot read your browsing history, bookmarks, passwords, or content on other websites. Its access is limited to the text fields on AI platforms where you might accidentally paste sensitive data.
What happens when I submit a false positive report
The report is sent to LeakSnitch servers with the detected text and platform context. No surrounding prompt content is included. The report helps improve detection accuracy. Patterns are updated periodically and distributed to all extension installations.
Is LeakSnitch free
LeakSnitch is free for individual users. The free tier includes full detection capabilities on all supported platforms and access to the personal dashboard with analytics. Enterprise features including organization management, custom rules, department policies, and SIEM integration require a paid plan.
Can I use LeakSnitch with my team
Yes. Enterprise plans support team based deployment with organization management, department grouping, and centralized policy enforcement. Users can be invited via email and assigned to departments. Administrators can view organization wide analytics and configure policies that apply to all members.
How do I update detection patterns
Detection patterns are updated automatically. The extension checks for new patterns every hour. Patterns are also bundled directly in the extension, so you are protected even if the pattern server is unreachable. No manual updates are required.
Does LeakSnitch work offline
Detection continues to work offline because all pattern matching runs locally in your browser. The bundled patterns ensure protection even without internet connectivity. Dashboard sync requires an internet connection to send detection metadata to the server.
What browsers are supported
LeakSnitch supports all Chromium based browsers including Google Chrome, Microsoft Edge, Brave, Opera, and Vivaldi. Firefox and Safari are not currently supported. Enterprise deployment via Group Policy is supported for managed environments.
What are the known limitations
LeakSnitch has several limitations worth noting. Only Chromium based browsers are supported; Firefox and Safari users cannot use the extension. The extension monitors 18 AI platforms and does not scan inputs on websites outside this list. File scanning supports PDFs, Word documents, Excel spreadsheets, and plain text files up to standard browser upload sizes; very large files may not be fully scanned. Detection runs on the last 4000 characters of each input to maintain performance. Dashboard features are only available with an active internet connection, though local detection continues offline. Enterprise features require a paid plan.
How do I contact support
If your question is not answered in this documentation, you can reach the LeakSnitch team through several channels. Email support is available at [email protected] for all inquiries including account issues, bug reports, and feature requests. Enterprise customers have access to dedicated support with prioritized response times. For security related issues, use the same email address with appropriate urgency markers. The LeakSnitch team aims to respond to all inquiries within one business day.
For complete details on how LeakSnitch handles your data and the terms governing your use of the service, please review the Privacy Policy and Terms of Service. For support inquiries, email [email protected].