zzejzznet / guard
LOCAL EVALUATION v0.1
EDGE LAYER

Checking the connection route…

BLUE TEAM / BROWSER INTELLIGENCE

Trust evidence.
Challenge assumptions.

Inspect browser consistency, compare an independent provider, and exercise real server gates. Each layer keeps its own verdict.

01 / LOCAL PROBES

Browser probes start automatically when this page loads. Reading the server’s full-report, mouse-recording and history settings…

LAYER 01INDEPENDENT ENGINE

zejzznet

Rendering consistency, API integrity and server-observed behavior.

STARTING SCANNo probability claim
Local probes run automatically after the session connects.
Canvas 2D
—
Offline Web Audio
—
WebGL rendering
—
API & context integrity
—
    LAYER 02EU REGION

    Fingerprint.com

    External Device Intelligence. Independent result, independent protected route.

    STARTINGProvider is not loaded

    Sends device signals to Fingerprint’s EU workspace using your public key. Runs automatically once per page load. Each run uses one identification from your Fingerprint allowance; the control panel can disable auto-run.

    Event
    —
    Visitor
    —
    Server verification
    —
    Provider history
    —

    A client-side visitor ID does not grant access.

    SEPARATE QUESTIONREADING RETENTION

    Seen before?

    A possible match to earlier browser reports. This is not proof of the same person.

    UNKNOWNAutomatic comparison starts with the browser scan
    Earliest matching visit
    —
    Latest matching visit
    —
    This visit last active
    —
    Matching history
    —
    Authenticated account review
    —

    No simulated accounts

    Simulation labels are separate from authenticated accounts. Matching accounts may share a device; this is a review signal, not proof of abuse.

      Common configurations and copied profiles can collide. Missing or unstable probes stay uncertain. Cookie deletion never restores an account lockout.

      02 / ENFORCEMENT WORKBENCH

      Try the protected flow.

      These actions hit the server. Low-risk browser reports pass automatically; suspicious or unavailable signals require verification. Checkout approval is separately bound to one frozen cart.

      Synthetic products only. No payments, card data, inventory, or live merchant systems.

      REQUEST JOURNAL
        Inspect residual statistics and collected evidence

        Standard deviation, skewness, kurtosis and lag correlation describe controlled readback residuals. They do not establish whether a user is malicious. Safari and other privacy browsers can deliberately change these outputs.

        No probe report yet.

        03 / EXPANDED MEASUREMENTS

        Inspect the whole report.

        62 general fields, 13 frame comparisons, full graphics and audio data, font signatures, browser capabilities, automation checks, and network observations. Missing APIs remain visible. This scan runs automatically after the baseline. Automation and strong context/header contradictions contribute to Layer 01. Other fields remain diagnostic; the comparisons below state which checks are enforced.

        Expanded measurements start after the baseline checks.

        Measurement groupStatus

        Network reputation

        Not measured

        Network observations
        No observations yet.
        Server comparisons and coverage
        No analysis yet.
        Full report returned by the server
        Use Inspect saved report to load full measured values. Download preserves all values without display truncation.

        04 / INTERACTION HISTORY

        Pointer and task typing recordings.

        Continuous raw movement, coalesced pointer samples, clicks, pressure, wheel, scroll, focus and viewport events from this page, plus keydown/up, input, editing and composition events in the exercise search field. Server timestamps accompany every batch. These measurements are diagnostic; absence of mouse input does not trigger a challenge.

        Starting with the session…

        Raw events returned by the server
        Move the pointer, click or scroll, then inspect the saved events.

        05 / INTERACTIVE BOT LAB

        Search. Compare. Select.

        A fresh synthetic marketplace exercise. Type into the search field, then select the listing matching the requested color and condition. The server records task timing and computes typing and pointer statistics from uploaded raw events.

        Connecting the exercise…

        Starting…

        Typing and pointer statistics
        Complete the exercise to see key hold/flight times, interval variability, skewness, kurtosis, path efficiency, speeds and pauses.

        This exercise is available to automated clients too. Completing it does not bypass a gate. Keyboard-only input, paste and touch are supported; behavioral measurements remain diagnostic until calibrated on real users.

        06 / ACTIVE CONSISTENCY & HARDWARE RESEARCH

        Fresh evidence through the interaction.

        Fresh server challenges run on load, first input, search and selection. Webdriver, mismatched main/worker environments and incorrect fresh control answers can contribute to Layer 01. Hardware timings, canvas residuals and behavioral distributions remain diagnostic.

        Connecting…

        WebGPU research

        WAITING

        LockedApart-inspired contention, pipeline cold/warm controls, and own-buffer memory access.

        Storage timing

        WAITING

        Bounded OPFS read/write/flush experiment. Supported browser APIs work across OS families; cached reads do not establish SSD identity.

        JA4 / TLS

        WAITING

          Latest interaction probe, full report
          Waiting for measurements.
          Hardware experiment, full report and candidates
          Waiting for measurements.

          INDEPENDENT PROVIDER / GOOGLE

          reCAPTCHA v3

          The score service used by Antcpt’s demo. Separate server verification and separate protected route. Runs on page load, search and selection when configured.

          WAITINGConnecting…
          Google score / 0 to 1
          —

          zejzznet never consumes this score. Google may return lower scores during testing.

          Verified provider result
          No result.

          Protected by reCAPTCHA; Google’s Privacy Policy and Terms of Service apply. Open the Antcpt reference demo.

          SEPARATE EXPERIMENT / ARXIV 2607.23389

          GPU render timing

          A bounded fragment-shader workload with synchronous readback, warm-up frames and raw timing distributions. This experiment is excluded from zejzznet and return-visit matching.

          WAITINGNo trained human/bot classifier or false-positive claim.
          Render samples and server statistics
          Waiting for measurements.