Protected by MagenTrust
    MAGEN
    MAGEN Docs

    Why MagenTrust

    MagenTrust adds continuous identity assurance for people, agents and machines to your applications. Instead of a one-time CAPTCHA, MagenTrust evaluates behavioral and entropy signals throughout the session and returns PASS, CHALLENGE or INTERRUPT before consequential actions, without adding user friction.

    Integrate MAGEN into your registration flows, checkout processes, or any user interaction where you need to verify authentic human behavior. The API works silently in the background, only escalating to visible challenges when risk levels warrant intervention.

    #How It Works

    MAGEN integrates seamlessly into your application's existing flow. The verification process happens invisibly in the background, with the SDK collecting behavioral signals while users interact naturally with your interface.

    Your AppFrontend
    1. Load SDK
    MAGEN SDKCollect Entropy
    2. Verify
    MAGEN APIAnalyze & Score
    3. Decision
    Your BackendValidate Token
    ① User loads your page, SDK initializes and starts passive collection
    ② SDK gathers mouse, keyboard, and touch entropy signals
    ③ Signals sent to MAGEN API, returns confidence score
    ④ Your backend verifies the session token server-side

    Data Flowing from MAGEN SDK

    The SDK passively collects behavioral entropy signals as users interact with your application. This data is compressed and sent to the API during verification.

    Mouse Events
    x, y coordinatesvelocityaccelerationclick patterns
    Keyboard Events
    key timingdwell timeflight timerhythm pattern
    Touch Events
    touch pointspressureswipe velocitygesture type
    Timing Metadata
    start timestampend timestampidle periodsinteraction gaps
    Sent to API
    // Entropy payload structure
    {
    "sessionId": "uuid",
    "challengeToken": "jwt",
    "entropy": {
    "mouse": [...events],
    "keyboard": [...events],
    "touch": [...events],
    "timing": {
    "start": 1704067200000,
    "end": 1704067205000
    }
    }
    }
    API Returns
    { "isHuman": true, "confidence": 0.95 }
    1Client-Side

    Install the MAGEN SDK and initialize it on page load. The SDK automatically begins collecting behavioral entropy.

    2Verification

    Before critical actions (signup, checkout), call verify() to submit entropy and receive a verdict token.

    3Server-Side

    Validate the token via GET /magen-verify-session to confirm the human classification before processing.

    #MagenTrust API Capabilities

    MAGEN API provides a comprehensive set of capabilities designed to help you keep identity assurance continuous across people, agents and machines while maintaining a seamless experience for legitimate users.

    Behavioral Fingerprinting

    Analyze mouse movements, keyboard patterns, scroll behavior, and touch gestures to create unique behavioral signatures.

    Entropy Signals

    Collect and evaluate randomness in user interactions to establish verifiable human presence.

    Real-time Scoring

    Get instant confidence scores and classifications within milliseconds, without blocking legitimate users.

    Adaptive Challenges

    Deploy graduated entropy challenges based on risk scores, from passive monitoring to active verification.

    #Behavioral Analysis

    MAGEN collects behavioral signals from user interactions to build a comprehensive profile of how they engage with your application. These signals include:

    • Mouse movement patterns, velocity, and acceleration
    • Keyboard timing, dwell time, and typing rhythm
    • Touch gestures, pressure sensitivity, and swipe patterns
    • Scroll behavior and viewport interactions
    • Device orientation and motion sensor data
    Passive Collection

    All behavioral data is collected passively in the background. Users never see a CAPTCHA or interruption unless the risk score triggers an escalation.

    #Entropy Challenges

    When passive analysis isn't sufficient to make a confident determination, MAGEN can deploy graduated entropy challenges. These challenges raise confidence in verifiable human presence while remaining invisible or minimally intrusive to humans.

    LevelChallenge TypeUser Impact
    1-2Extended observation periodNone
    3-4Timing jitter, UI randomizationImperceptible
    5-7Proof-of-work, interaction requirementsMinor delay
    8-10Visible challenge, manual verificationUser action required

    #Real-time Decisions

    MAGEN processes behavioral signals and returns a decision in real-time, typically within 50-100ms. Each response includes:

    Decision
    PASS / CHALLENGE / INTERRUPT
    Confidence
    0.0 - 1.0 score
    Risk Level
    low / medium / high

    #Session Management

    MAGEN uses secure, token-based sessions to track verification state across user interactions. Sessions are ephemeral by design—they expire after 5 minutes of inactivity and cannot be reused once a decision is finalized.

    JWT-based Tokens

    All session tokens are signed JWTs that can be verified client-side or server-side. Tokens include session ID, classification lane, confidence score, and expiration time.

    #Easy Integration

    MAGEN provides official SDKs for popular platforms, making integration straightforward. A typical implementation requires just a few lines of code.

    JavaScriptReactReact NativePythonGoRuby

    #Use Cases

    MAGEN is designed to protect critical user interactions across your application:

    Account Registration
    Confirm a real person is present when accounts are created
    Payment Flows
    Require verified presence before high-value checkout actions
    Form Submissions
    Confirm who is submitting before forms are accepted
    Login Protection
    Keep assurance continuous after sign-in, not just at login
    API Rate Limiting
    Tell verified people, authorized agents and unverified callers apart
    Content Access
    Prevent content scraping and unauthorized downloads

    Ready to protect your application?

    Get started with MAGEN in minutes with our quick start guide.