{
  "report": "Browser Busters: browser security briefings and owned-application checks",
  "reportRelease": 14,
  "scope": {
    "applications": [
      "/supplier-file-converter/",
      "/margin-guard/"
    ],
    "environment": "Windows; fresh automation profiles; synthetic files",
    "widthsCssPixels": [
      320,
      390,
      768,
      1440
    ],
    "states": [
      "initial screen",
      "sample conversion or expanded audit",
      "explicit approval and CSV download"
    ],
    "limits": [
      "Curated vendor advisories and bounded authorized application tests; not an exhaustive vulnerability catalog or browser-engine penetration test.",
      "No browser exploit, third-party scan or personal-profile/extension inspection was performed.",
      "Desktop CSS viewport emulation does not establish physical mobile-browser behavior.",
      "Functional checks and vendor disclosures do not determine a device's patch status or local compromise.",
      "Checked versions and scenarios do not establish all-version compatibility."
    ],
    "applicationReleases": {
      "/supplier-file-converter/": 2,
      "/margin-guard/": 3
    },
    "coverageNote": "The original two-application coverage and release numbers are retained as historical evidence for Releases 1–4. startupRegressionSummary records the separate four-application startup repair."
  },
  "observations": [
    {
      "browser": "Microsoft Edge",
      "engine": "Chromium",
      "version": "157.0.4341.46",
      "result": "passed",
      "evidence": "Four-width app checks plus the live supplier-to-margin workflow."
    },
    {
      "browser": "Google Chrome",
      "engine": "Chromium",
      "version": "156.0.8078.48",
      "result": "passed",
      "applicationViewportWorkflows": 8
    },
    {
      "browser": "Playwright WebKit on Windows",
      "engine": "WebKit",
      "version": "26.5",
      "result": "passed",
      "applicationViewportWorkflows": 8,
      "qualification": "Workflow console checks exclude screenshot instrumentation. Separate captured images were reviewed. This is not branded Safari."
    },
    {
      "browser": "Firefox / Gecko",
      "result": "untested",
      "reason": "The official test-runtime installer failed; no runnable Firefox result was obtained."
    },
    {
      "browser": "Physical Safari / iPhone / Android",
      "result": "untested",
      "reason": "No physical-device check was performed."
    },
    {
      "browser": "Brave / Opera / other browsers, extensions and embedded webviews",
      "result": "untested",
      "reason": "A shared engine does not establish these product and session behaviors."
    }
  ],
  "findings": [
    {
      "id": "FMP-UNICODE-001",
      "title": "How a split Unicode character cleared visible results",
      "classification": "Owned application reliability; not a browser-engine vulnerability",
      "application": "/find-my-people",
      "applicationRelease": 7,
      "prerequisites": "The demonstrated source path begins with a valid business name containing a character represented by a UTF-16 pair across the producer’s 180-code-unit cutoff. The fictional record passes the relevant source category and location filters and has no usable website, so the browser builds a Find contact link. Our own normalizer splits the character. We did not establish that a real provider currently supplies such a record or that any attacker controls it. The seeded saved-list case is a separate prerequisite, not proof that the baseline UI can normally save the already-failing record.",
      "observed": "My People showed one fictional search result. A later listing passed the string checks, but its fallback contact-link construction threw an encoding error and left zero visible result cards. The prior saved shortlist remained intact. A separate startup test with an already-saved malformed name showed zero saved cards and did not initialize history; the original storage bytes remained present.",
      "transition": "Valid source Unicode → bounded JSON decoding → source normalization truncates the name → accepted record and serialized cache payload → browser parses and validates the string type → replacement commits → list clears → fallback URL encoding throws. The independent producer-path check used the actual functions with an injected Response and mocked cache, not a deployed-worker request.",
      "repair": "Build the fixed HTTPS contact-search URL with URL and URLSearchParams. An unpaired surrogate becomes a replacement character only in that outgoing query. Keep the original listing name, identifier and saved record unchanged. The same card renderer now works during search and saved-list restoration. This is a consumer repair: upstream truncation remains unchanged, and the page does not claim that all Unicode source handling has been corrected.",
      "controls": "The original review recorded 14 Unicode-related observations, including six failures. An independent review used the actual producer output at three adjacent cutoff positions: only the pair split by the cutoff failed. Valid ASCII, Arabic, CJK, combining marks and complete astral characters worked. A safe Website bypassed the failing fallback. Invalid raw UTF-8 was replaced during decoding and did not reproduce the escaped-surrogate failure. The repaired run passed all 14 Unicode-related observations, including ten actual Save-and-reload journeys; saved bytes were preserved.",
      "caseUrl": "https://tenthirtytwo.com/browser-audit/unicode-contact-recovery/",
      "limits": "Fictional local inputs, fresh profiles and named engines only. No production occurrence, customer loss, incident frequency, attacker access, browser-engine flaw or security compromise established. Physical Safari/iPhone/Android, native keyboards and Firefox untested. Exact source and failed/repaired receipts retained privately.",
      "baseline": {
        "observations": 14,
        "environments": [
          {
            "browser": "Microsoft Edge",
            "version": "157.0.4341.46"
          },
          {
            "browser": "Playwright WebKit on Windows",
            "version": "26.5",
            "qualification": "Not physical Safari or a mobile device."
          }
        ],
        "replacementFailures": 4,
        "seededStartupFailures": 2,
        "priorSavedBytesPreserved": true,
        "receipt": "deep-input-baseline-01"
      },
      "independent": {
        "receipt": "independent-disproof-02",
        "sourcePairPositions": [
          178,
          179,
          180
        ],
        "sourceNamesWellFormed": true,
        "onlyCrossingPairFailed": true,
        "actualProducerPath": true,
        "mockedCacheOnly": true,
        "rawInvalidUtf8ControlRendered": true,
        "productionOccurrenceEstablished": false
      },
      "repaired": {
        "observations": 14,
        "passed": 14,
        "saveReloadJourneys": 10,
        "sourceRecordPreserved": true,
        "receipt": "deep-input-fixed-01"
      },
      "unchangedScope": "The upstream code-unit truncation remains; this release repairs client contact-link encoding and the shared card renderer."
    },
    {
      "id": "FMP-METADATA-001",
      "title": "When nested source metadata interrupted a listing",
      "classification": "Owned application reliability; not a browser-engine vulnerability",
      "application": "/find-my-people",
      "applicationRelease": 7,
      "prerequisites": "The candidate requires malformed nested location evidence for a supported source, with the surrounding attribution fields matching that source. No native producer was shown to create this object shape, and no production injection or attacker access was established. The same unsupported identifier with an unknown source key was ignored in the independent control; a canonical string identifier retained legitimate attribution.",
      "observed": "A fictional home-health replacement had the expected listing identity, state, supported source, method, URL and date. Its nested record identifier was an object instead of a string, with non-callable conversion fields. The source-identifier check tried to convert it and threw, leaving zero visible replacement cards. The prior saved shortlist stayed intact. A separately seeded saved-list test also interrupted startup while retaining the original storage bytes.",
      "transition": "Receive → parse → validate ordinary listing fields → commit replacement → clear visible list → inspect optional supported-source metadata → implicit object-to-string conversion throws. On seeded startup, the same rendering step interrupts saved-list display before navigation is initialized.",
      "repair": "Require primitive strings before validating location-evidence record IDs and dates. Apply the same check to secondary Ohio and Illinois location-source IDs and the Illinois checked date. Ignore unsupported optional metadata while retaining the underlying listing and its original saved record. Existing source, method, URL, identity and calendar checks still apply; valid source attribution remains visible.",
      "controls": "The initial review recorded eight metadata-related observations, including four replacement/startup failures. Independent checks reproduced the supported-source failure in both engines and retained cards for valid string and unknown-source controls. All 14 repaired metadata observations passed, including twelve actual Save-and-reload journeys and adjacent malformed date or secondary-ID cases. Unit checks preserve valid Texas, Ohio and Illinois attribution while rejecting wrong types and existing mismatched evidence.",
      "caseUrl": "https://tenthirtytwo.com/browser-audit/nested-source-metadata/",
      "limits": "Fictional local inputs, fresh profiles and named engines only. No production occurrence, customer loss, incident frequency, attacker access, browser-engine flaw or security compromise established. Physical Safari/iPhone/Android, native keyboards and Firefox untested. Exact source and failed/repaired receipts retained privately.",
      "baseline": {
        "observations": 8,
        "environments": [
          {
            "browser": "Microsoft Edge",
            "version": "157.0.4341.46"
          },
          {
            "browser": "Playwright WebKit on Windows",
            "version": "26.5",
            "qualification": "Not physical Safari or a mobile device."
          }
        ],
        "replacementFailures": 2,
        "seededStartupFailures": 2,
        "priorSavedBytesPreserved": true,
        "receipt": "deep-input-baseline-01"
      },
      "independent": {
        "receipt": "independent-disproof-02",
        "supportedSourceReproducedBothEngines": true,
        "unknownSourceControlRendered": true,
        "nativeProducerOfMalformedObjectEstablished": false
      },
      "repaired": {
        "observations": 14,
        "passed": 14,
        "saveReloadJourneys": 12,
        "invalidAttributionIgnored": true,
        "sourceRecordPreserved": true,
        "receipt": "deep-input-fixed-01"
      },
      "unitControls": [
        "Valid Texas, Ohio and Illinois source attribution",
        "Primitive-string IDs and dates",
        "Calendar and source mismatch rejection",
        "Adjacent secondary location IDs and checked dates"
      ]
    },
    {
      "id": "FMP-CATALOG-001",
      "classification": "Owned application reliability; not a browser-engine vulnerability",
      "application": "/find-my-people",
      "applicationRelease": 6,
      "title": "When broken suggestions stop a typed search",
      "prerequisites": "The browser receives an unexpected catalog shape, such as an object instead of an array, a null member or object-valued aliases. The test intercepts a response locally; it does not establish that a visitor or attacker can alter the production catalog.",
      "observed": "A fresh My People page received a malformed occupation catalog. The page said suggestions were unavailable and that a service could still be typed, but typing or submitting then failed before a search request was made. A valid catalog and a valid empty catalog each allowed one fictional search and one visible result.",
      "transition": "Receive → parse → commit categories → later input uses the committed value → exception. The fallback message did not undo the earlier assignment.",
      "repair": "Read the response within a 1 MiB byte limit, validate and detach the full catalog, and prepare its controls before committing it. Reject malformed or oversized suggestions while retaining the usable category state. A valid empty catalog remains a supported state, so typed search still works.",
      "controls": "The unchanged-source run covered eight catalog shapes in each engine. Positive controls were a valid catalog and a valid empty catalog. Repaired cases require a completed typed search and no uncaught page error. Unit checks also exercise the actual 1,027-entry catalog, duplicate slugs, UTF-8 byte limits and cancellation of oversized streams.",
      "limits": "Owned local copies, fictional records and fresh profiles. No malformed response sent to production, remote attacker path, measured customer loss, incident frequency, browser-engine flaw, prototype pollution or account-isolation guarantee. Firefox, physical Safari/iPhone/Android and native mobile keyboards remain untested. Exact source hashes, original failures and completed repaired receipts are retained privately.",
      "caseUrl": "https://tenthirtytwo.com/browser-audit/category-catalog-recovery/",
      "baseline": {
        "cases": 16,
        "environments": [
          {
            "browser": "Microsoft Edge",
            "version": "157.0.4341.46"
          },
          {
            "browser": "Playwright WebKit on Windows",
            "version": "26.5",
            "qualification": "Patched test build, not physical Safari or a mobile device."
          }
        ],
        "blockedTypedSearchCases": 8,
        "validAndEmptyControls": 4,
        "otherNonFailingCases": 4,
        "receipt": "catalog-baseline-01"
      },
      "repaired": {
        "cases": 16,
        "passed": 16,
        "completedTypedSearches": 16,
        "visibleResultEach": 1,
        "pageErrors": 0,
        "unexpectedExternalRequests": 0,
        "receipt": "catalog-fixed-02"
      },
      "additionalUnitBoundaries": {
        "actualCatalogEntries": 1027,
        "receivedByteLimit": 1048576,
        "oversizedStreamCanceled": true,
        "declaredLengthTrusted": false,
        "browserOversizedStreamScenarioClaimed": false
      }
    },
    {
      "id": "FMP-RESTORE-001",
      "classification": "Owned application reliability; not a browser-engine vulnerability",
      "application": "/find-my-people",
      "applicationRelease": 6,
      "title": "Recovering a saved list without overwriting it",
      "prerequisites": "Malformed data must already be present in this browser’s saved list, arrive through its storage event, or exist in a history entry. The tests seed fictional local state. No remote path for altering a visitor’s storage or history was demonstrated.",
      "observed": "A synthetic saved list contained one valid person and a second record with a numeric phone or object-shaped listing type. Both records passed the old basic identity check. Rendering stopped after one card, while the saved count showed two; startup navigation did not initialize. A malformed cross-tab replacement and a pre-existing malformed history entry also interrupted restoration. Original local-storage bytes were not deleted.",
      "transition": "Read or restore → basic id/name check → commit saved/history state → render optional fields → exception. Validating only identity did not establish that the record could be rendered.",
      "repair": "Use a shared listing contract before rendering or restoring. At startup, show usable recovered entries while retaining the original storage bytes. Pause mutations when recovery is needed; explicitly clearing the entire saved list is a deliberate reset. Reject a malformed cross-tab replacement before changing the committed in-tab list. Reject malformed history snapshots before applying them. A storage event does not undo another tab’s write: the repair preserves this tab’s committed view, not an atomic cross-tab transaction.",
      "controls": "Valid and empty stored lists remain usable. Repaired checks cover mixed and all-invalid records, preservation of raw storage, blocked implicit overwrites, explicit clearing, cross-tab rejection, malformed history and ordinary filters with Back, Forward and refresh. These are local-state recovery checks, not account authorization tests.",
      "limits": "Owned local copies, fictional records and fresh profiles. No malformed response sent to production, remote attacker path, measured customer loss, incident frequency, browser-engine flaw, prototype pollution or account-isolation guarantee. Firefox, physical Safari/iPhone/Android and native mobile keyboards remain untested. Exact source hashes, original failures and completed repaired receipts are retained privately.",
      "caseUrl": "https://tenthirtytwo.com/browser-audit/saved-list-recovery/",
      "baseline": {
        "observations": 18,
        "environments": [
          {
            "browser": "Microsoft Edge",
            "version": "157.0.4341.46"
          },
          {
            "browser": "Playwright WebKit on Windows",
            "version": "26.5",
            "qualification": "Patched test build, not physical Safari or a mobile device."
          }
        ],
        "receipt": "saved-state-baseline-03",
        "savedBytesDeleted": false,
        "twoEngineHistoryFinding": "Malformed preexisting history entry; not the engine-dependent immediately-before-reload corruption probe."
      },
      "repaired": {
        "observations": 20,
        "passed": 20,
        "receipt": "saved-state-fixed-02",
        "rawStartupStoragePreserved": true,
        "implicitOverwriteBlocked": true,
        "deliberateClearSupported": true,
        "allInvalidRecovery": true,
        "malformedCrossTabCandidateRejectedBeforeCommit": true,
        "malformedHistoryRejectedBeforeApply": true,
        "ordinaryFiltersBackForwardRefreshPassed": true,
        "scope": "The other tab’s storage write remains; this tab preserves its committed view. This is not a transactional storage or account-isolation claim."
      }
    },
    {
      "id": "FMP-SOURCE-001",
      "classification": "Owned application reliability; not a browser-engine vulnerability",
      "application": "/find-my-people",
      "applicationRelease": 6,
      "title": "When an inherited source name cleared visible results",
      "prerequisites": "A controlled replacement response contains a home-health listing ID and category, a locationEvidence object whose sourceId is an inherited object-property name, no listing state, and no location-evidence method or URL. The comparison uses an ordinary unknown source name with the same absent fields. No production injection, attacker access or real customer incident was established.",
      "observed": "My People showed one fictional result and saved it. A replacement home-health record used an inherited object-property name as its location source, with the listing state and the location-evidence method and URL absent. The plain object lookup returned an inherited value, and rendering later tried to use a missing validation rule. The replacement left zero visible cards. The saved shortlist remained intact.",
      "transition": "Receive → basic listing validation → commit replacement → lookup source name on a plain object → render against an inherited value → exception. This was a lookup error, not demonstrated prototype pollution or an engine isolation failure.",
      "repair": "Require the source label to be an own property of the application’s fixed source registry before consulting its validation rules. Unsupported metadata is ignored. The listing can still render, and supported source attribution is preserved.",
      "controls": "Three inherited names and one ordinary unknown-source control were checked in each engine on exact baseline source. The six inherited-name cases changed one visible result to zero; both unknown-source controls retained one. Repaired checks require one visible result and preserved saved state for every case.",
      "limits": "Owned local copies, fictional records and fresh profiles. No malformed response sent to production, remote attacker path, measured customer loss, incident frequency, browser-engine flaw, prototype pollution or account-isolation guarantee. Firefox, physical Safari/iPhone/Android and native mobile keyboards remain untested. Exact source hashes, original failures and completed repaired receipts are retained privately.",
      "caseUrl": "https://tenthirtytwo.com/browser-audit/inherited-source-name/",
      "baseline": {
        "cases": 8,
        "environments": [
          {
            "browser": "Microsoft Edge",
            "version": "157.0.4341.46"
          },
          {
            "browser": "Playwright WebKit on Windows",
            "version": "26.5",
            "qualification": "Patched test build, not physical Safari or a mobile device."
          }
        ],
        "inheritedNameFailures": 6,
        "ordinaryUnknownNameControls": 2,
        "displayedBefore": 1,
        "displayedAfterInheritedName": 0,
        "savedPreservedInAllCases": true,
        "receipt": "source-key-baseline-01"
      },
      "repaired": {
        "cases": 8,
        "passed": 8,
        "visibleResultEach": 1,
        "savedPreservedInAllCases": true,
        "pageErrors": 0,
        "unexpectedExternalRequests": 0,
        "receipt": "source-key-fixed-02"
      },
      "additionalUnitBoundaries": {
        "supportedTexasAttributionPreserved": true,
        "mismatchedEvidenceRejected": true
      },
      "notDemonstrated": [
        "prototype pollution",
        "attacker control over production responses",
        "browser-engine isolation failure"
      ]
    },
    {
      "id": "FMP-RESPONSE-001",
      "classification": "Owned application reliability; not a browser-engine vulnerability",
      "severity": "Low; lost visible search state, without demonstrated security compromise",
      "application": "/find-my-people",
      "applicationRelease": 5,
      "prerequisites": "A visitor has completed search results, then a later HTTP 200 response contains an unusable listing. A controlled fixture used a null listing name.",
      "observed": "Twelve displayed fictional listings became zero and history navigation became unavailable. The saved shortlist remained intact.",
      "repair": "Validate the candidate result collection, listing identity and optional text fields, directory links and serialized size before replacing committed data. Rejected candidates retain previous results, filters and navigation; valid empty results still replace the prior search.",
      "baseline": {
        "cases": 2,
        "environments": [
          "Microsoft Edge 157.0.4341.46",
          "Playwright WebKit on Windows 26.5"
        ],
        "widthCssPixels": 390,
        "displayedBefore": 12,
        "displayedAfterMalformedReplacement": 0,
        "savedPreserved": true
      },
      "repaired": {
        "cases": 2,
        "passed": 2,
        "displayedAfterMalformedReplacement": 12,
        "status": "Search returned incomplete listings. Your previous results are still available. Please try again.",
        "savedPreserved": true
      },
      "regression": {
        "cases": 6,
        "passed": 6,
        "scope": "Five malformed response shapes preserve committed results, filters and history; a valid empty response replaces prior results."
      },
      "controls": {
        "ordinaryHttp503PreservedResultsBeforeAndAfter": true,
        "markupShapedNameRemainedText": true,
        "unsafeWebsiteLinks": 0,
        "pageErrors": 0,
        "unexpectedExternalRequests": 0
      },
      "limits": "Owned local source, fresh profiles and fictional intercepted responses. No production fault injection, customer incident frequency, actual customer loss, demonstrated attacker-controlled path or browser-engine flaw. Windows Playwright WebKit is not Apple Safari. Firefox, physical iPhone/Android and native keyboards are untested."
    },
    {
      "id": "OCA-RECOVERY-001",
      "classification": "Owned application reliability; not a browser-engine vulnerability",
      "severity": "Low; local review-work loss, not a demonstrated security compromise",
      "application": "/order-confirmation-auditor/",
      "prerequisites": "A visitor has a comparison with review notes, then asks to load another fictional sample. That sample request fails before the pair is ready.",
      "observed": "The old implementation cleared the prior pair, result and review notes before the request completed. A failed replacement left no in-tab comparison to export. Original files and previously downloaded reports were not deleted.",
      "controls": "Changing only the sample selector retained a byte-identical reviewed report; a completed explicit replacement correctly began a new review.",
      "repair": "Prepare and validate the candidate pair separately. Keep the committed workspace while it loads, block compare/review editing/export during the pending attempt, and replace both sources only on current-candidate success. Current failure keeps prior annotations; later input actions cancel pending work without resurrecting old reviews.",
      "applicationRelease": 3,
      "baseline": {
        "failedLoadReproductions": 2,
        "environments": [
          "Microsoft Edge 157.0.4341.46",
          "Playwright WebKit on Windows 26.5"
        ],
        "additionalIndependentEdgeReproduction": 1
      },
      "regression": {
        "cases": 23,
        "passed": 23,
        "scope": "Local synthetic source/review transitions, malformed samples, pending exports and canceled or superseded requests; one Edge case checks interrupted selected-file resumption."
      },
      "limits": "These checks do not establish incident frequency, actual customer loss, Firefox, physical Safari/mobile compatibility, browser patch status or a browser-engine flaw. Detached-control dispatches are controlled robustness probes, not evidence of an attacker capability."
    },
    {
      "id": "APP-STARTUP-001",
      "classification": "Confirmed owned-application readiness defect; not a browser-engine vulnerability",
      "severity": "Low: lost interaction and misleading file-selection state under delayed initialization",
      "applications": [
        "/margin-guard/",
        "/supplier-file-converter/",
        "/supplier-quote-comparator/",
        "/order-confirmation-auditor/"
      ],
      "prerequisite": "The visitor acts after HTML controls appear but before the application module finishes initialization.",
      "finding": "Three sample buttons and four file inputs accepted early actions without updating application state. File inputs held one synthetic file while application status did not reflect the selected file.",
      "control": "Margin Guard already disabled its demo button until ready; that negative control passed in all three tested environments. All tested controls worked after initialization.",
      "repair": "Initial intake and policy controls remain disabled until startup completes; conditional compare and export controls retain their prerequisites. Visible help remains available if startup fails or JavaScript is unavailable.",
      "status": "Corrected in Margin Guard Release 6, Supplier File Converter Release 3, Supplier Quote Comparator Release 2 and Order Confirmation Auditor Release 2.",
      "limits": "A controlled delay demonstrates a possible interaction window; it does not measure how often visitors encounter it. File selection was supplied through browser automation, not a physical operating-system picker. No data disclosure, browser-engine exploit, account compromise or native mobile behavior is established."
    },
    {
      "id": "SFC-IDENTITY-001",
      "application": "Supplier File Converter",
      "title": "A formatting-character blacklist missed an invisible SKU",
      "severity": "Low, conditional application data-integrity impact",
      "prerequisites": "The visitor selects and maps a crafted supplier file, validates it, explicitly reviews the rows and prepares the import CSV.",
      "observed": "A SKU consisting only of HANGUL FILLER passed validation and was exported unchanged through the actual review and download flow. Existing rejection of ZERO WIDTH SPACE and formula prefixes did not catch this case.",
      "correction": "Require visible identity content in addition to the formatting-control checks. Keep legitimate Arabic/CJK text, leading zeroes and joiners within visible identifiers intact; preserve invalid source evidence for correction.",
      "verification": "Canonical core checks passed 12 of 12 after new regressions failed on the original source. The staged browser blocked all four negative-control rows from import. Six valid identifiers passed the Converter-to-Margin workflow without changing their exact text.",
      "status": "Corrected in the application release named in scope.applicationReleases; original failures and targeted regressions reproduced in the controlled test environment.",
      "limits": "No inventory import or third-party spreadsheet execution was attempted. No data disclosure or unexpected network request was observed. The finding concerns accepted identity data, not a browser-engine exploit.",
      "classification": "Owned application data-integrity issue corrected"
    },
    {
      "id": "MG-IDENTITY-002",
      "application": "Margin Guard",
      "title": "An invisible SKU could reach an approved price-change export",
      "severity": "Low, conditional application data-integrity impact",
      "prerequisites": "The visitor selects matching cost and selling-price files containing the crafted SKU, runs the audit and explicitly approves an export.",
      "observed": "A SKU consisting only of ZERO WIDTH SPACE was displayed without a visible part name, classified below target and exported unchanged after manual approval. Separate core checks also accepted other invisible-only identifiers and directional formatting.",
      "correction": "Require visible identity content and reject unsafe controls or directional formatting before permitting pricing advice or export. Preserve the original source value; do not silently strip it or invent a replacement.",
      "verification": "Canonical core checks passed 28 of 28 after the regression failed on the original source. The staged browser blocked approval and export for all three negative-control rows. Six valid multilingual or leading-zero identifiers retained exact identity through the paired tools and changes/rollback exports.",
      "status": "Corrected in the application release named in scope.applicationReleases; original failures and targeted regressions reproduced in the controlled test environment.",
      "limits": "No inventory update, monetary loss, data disclosure, unexpected network request or script execution was observed. This is not a browser-engine vulnerability or a claim that every Unicode identifier is safe.",
      "classification": "Owned application data-integrity issue corrected"
    },
    {
      "id": "MG-RESOURCE-001",
      "classification": "Owned application issue corrected",
      "severity": "Low; conditional availability impact",
      "prerequisite": "A visitor selects a crafted or unusually wide local CSV and runs Margin Guard.",
      "observed": {
        "costFileBytes": 226863,
        "dataRows": 500,
        "costColumns": 410,
        "totalDomElements": 437704,
        "closedDetailElements": 425524,
        "mainThreadTaskMilliseconds": 1099
      },
      "effect": "Temporary interaction blocking in the controlled test. No crash, code execution or data disclosure observed.",
      "fix": "100-column parser limit and deferred construction of detailed source evidence until that row opens.",
      "regression": {
        "rejectedColumns": 410,
        "acceptedColumnBoundary": 100,
        "acceptedDataRows": 500,
        "beforeBoundaryDomElements": 127704,
        "afterBoundaryDomElements": 12705,
        "initialEvidenceBodies": 0,
        "oneOpenedRowAddedElements": 230,
        "sourceFieldsInOpenedRow": 107
      },
      "certainty": "High for the reproduced app behavior and regression cases. Counts and timings are environment-specific, not a universal denial-of-service or performance claim."
    },
    {
      "id": "negative-controls",
      "classification": "Bounded negative application tests",
      "checks": [
        "Markup-shaped input remained literal text without script execution or injected elements.",
        "Prototype-shaped input did not modify object prototypes.",
        "Delayed old file reads and hashes could not restore cleared results or approvals.",
        "Withdrawing export approval revoked the prepared download Blob URL.",
        "Formula-risk identifiers were blocked from approved imports.",
        "Both full applications rejected same-origin framing.",
        "The explicit calculator embed worked same-origin and rejected a different origin.",
        "Synthetic workflows made no external data requests and left no localStorage or sessionStorage entries."
      ],
      "limits": "Fresh Edge automation contexts, owned local apps and finite synthetic cases. No claim that every race, injection, extension or session is safe."
    },
    {
      "id": "coverage",
      "classification": "Open coverage gap",
      "finding": "Physical mobile browsers, branded Safari, Firefox and other named products remain untested.",
      "next": "Obtain explicit real-device and Gecko test evidence before expanding compatibility claims."
    },
    {
      "id": "port",
      "classification": "Observed test-environment behavior",
      "finding": "WebKit rejected local port 4190 before application loading. The same app files passed on port 4174.",
      "action": "Use an allowed development port; browser security settings were not bypassed.",
      "reference": "https://raw.githubusercontent.com/WebKit/WebKit/main/Source/WTF/wtf/URL.cpp",
      "locator": "portAllowed; blockedPortList; entry 4190"
    },
    {
      "id": "capture",
      "classification": "Observed test-tool behavior",
      "finding": "In the tested WebKit runtime, screenshot capture produced an inline-style CSP console error. A control run without screenshot capture had clean application console checks.",
      "action": "Keep the strict site CSP. Separate capture-induced messages from application failures.",
      "certainty": "High for these controlled runs; no claim about every Playwright or WebKit version."
    },
    {
      "id": "touch",
      "classification": "Application issue corrected",
      "finding": "A companion link initially measured approximately 42.09 CSS pixels high at a 320-pixel viewport.",
      "action": "Added a 44-pixel minimum target; passing measurements were verified after the change."
    },
    {
      "id": "checksum",
      "classification": "File-provenance check",
      "finding": "The live supplier-to-margin round trip preserved SKU 00123 and matched the uploaded UTF-8 BOM file bytes to its displayed SHA-256 checksum.",
      "action": "Preserve the BOM when decoding uploaded text before hashing."
    }
  ],
  "syntheticCase": {
    "sourceRows": 11,
    "acceptedRows": 4,
    "blockedRows": 7,
    "downstreamAuditedRows": 4,
    "downstreamBlockedRows": 0
  },
  "references": [
    {
      "title": "Playwright browser documentation",
      "url": "https://playwright.dev/docs/browsers#webkit",
      "supports": "Playwright WebKit uses a patched runtime and does not establish branded Safari coverage."
    },
    {
      "title": "WebKit URL.cpp",
      "url": "https://raw.githubusercontent.com/WebKit/WebKit/main/Source/WTF/wtf/URL.cpp",
      "supports": "Port 4190 is included in the source blocked-port list."
    }
  ],
  "certainty": "Direct observations support the named versions, source and scenarios. Vendor disclosures support attributed claims, with branch/platform boundaries and source discrepancies retained. Untested security behavior remains unknown.",
  "vendorResearch": {
    "title": "Browser security: six evidence-backed findings",
    "mode": "curated primary-source evidence landscape and release proofing",
    "retrieved": "2026-11-10",
    "scope": "Published vendor/CNA/CISA records and architecture documentation, not original vulnerability discoveries or device-risk findings.",
    "count": {
      "notes": 6,
      "advisoryCaseGroups": 5,
      "architectureBoundaryNotes": 1
    },
    "findings": [
      {
        "id": "chrome-xml",
        "title": "Chrome XML fix has a documented Extended Stable backport",
        "kind": "published-vendor-advisory",
        "cves": [
          "CVE-2026-125350"
        ],
        "severity": "High (Google)",
        "affected": {
          "product": "Google Chrome",
          "platforms": [
            "Windows",
            "macOS",
            "Linux"
          ],
          "reportedRange": "CNA description: before 157.0.8103.50; this generic cutoff does not encode the explicit Extended Stable backport.",
          "completeHistoryVerified": false
        },
        "fixed": [
          {
            "product": "Chrome Stable",
            "platform": "Windows/macOS",
            "versions": [
              "157.0.8103.50",
              "157.0.8103.51"
            ],
            "releaseDate": "2026-11-09"
          },
          {
            "product": "Chrome Stable",
            "platform": "Linux",
            "versions": [
              "157.0.8103.50"
            ],
            "releaseDate": "2026-11-09"
          },
          {
            "product": "Chrome Extended Stable",
            "platform": "Windows/macOS",
            "versions": [
              "156.0.8078.96"
            ],
            "releaseDate": "2026-11-09"
          }
        ],
        "exploitation": {
          "vendor": "Google reports an exploit in the wild.",
          "kev": {
            "listed": true,
            "dateAdded": "2026-11-10"
          }
        },
        "interpretation": "CNA impact is code execution inside the sandbox. Use the correct release branch; a universal below-157 rule would misclassify the documented backport.",
        "certainty": {
          "reportedFacts": "high",
          "currentInstalledRisk": "not determined",
          "basis": "Google official release text retrieved through the search index, plus a fresh CVE API record and CISA catalog. Stable and Extended Stable branches checked separately."
        },
        "limits": [
          "No local exploitability or compromise conclusion.",
          "A release is not proof of installation or restart.",
          "No Edge fix is inferred from Chrome's build number."
        ],
        "sources": [
          {
            "id": "S01",
            "title": "Stable Channel Update for Desktop",
            "url": "https://chromereleases.googleblog.com/2026/11/stable-channel-update-for-desktop_018223192.html",
            "locator": "Opening release paragraph; CVE-2026-125350 entry; exploitation statement"
          },
          {
            "id": "S02",
            "title": "Extended Stable Update for Desktop",
            "url": "https://chromereleases.googleblog.com/2026/11/extended-stable-update-for-desktop_0910980127.html",
            "locator": "Opening version paragraph and CVE-2026-125350 list item"
          },
          {
            "id": "S03",
            "title": "CVE-2026-125350",
            "url": "https://cveawg.mitre.org/api/cve/CVE-2026-125350",
            "locator": "containers.cna.descriptions; containers.cna.affected; cveMetadata"
          },
          {
            "id": "S04",
            "title": "Known Exploited Vulnerabilities catalog",
            "url": "https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json",
            "locator": "vulnerabilities entry matching the named CVE"
          }
        ]
      },
      {
        "id": "chromium-v8-edge",
        "title": "One Chromium issue has separate Chrome and Edge fix entries",
        "kind": "published-vendor-advisory",
        "cves": [
          "CVE-2026-87491"
        ],
        "severity": "Medium (Google)",
        "affected": {
          "product": "Chrome/Chromium V8; Edge applicability confirmed by Microsoft fix entry",
          "reportedRange": "Chrome CNA description: before 153.0.8010.36. Complete Edge affected interval not verified.",
          "completeHistoryVerified": false
        },
        "fixed": [
          {
            "product": "Chrome Stable",
            "platform": "Linux",
            "versions": [
              "153.0.8010.36"
            ],
            "releaseDate": "2026-09-08"
          },
          {
            "product": "Chrome Stable",
            "platform": "Windows/macOS",
            "versions": [
              "153.0.8010.36",
              "153.0.8010.37"
            ],
            "releaseDate": "2026-09-08"
          },
          {
            "product": "Microsoft Edge Stable",
            "platform": "Stable desktop release entry; do not transfer to mobile",
            "versions": [
              "152.0.4191.66"
            ],
            "releaseDate": "2026-09-04"
          }
        ],
        "exploitation": {
          "vendor": "Google reports exploitation; Microsoft relays Chromium's statement in its named fix entry.",
          "kev": {
            "listed": true,
            "dateAdded": "2026-09-09"
          }
        },
        "interpretation": "Severity label and observed exploitation answer different questions. The described V8 effect remains inside the sandbox; this is one shared CVE.",
        "certainty": {
          "reportedFacts": "high",
          "currentInstalledRisk": "not determined",
          "basis": "Google official release text retrieved through the search index, Microsoft direct release notes, plus a fresh CVE API record and CISA catalog. Product build schemes retained."
        },
        "limits": [
          "No independent discoveries implied by two product entries.",
          "Vendor entry dates are not a reconstructed attack timeline.",
          "Edge and Chrome build schemes differ."
        ],
        "sources": [
          {
            "id": "S05",
            "title": "Release notes for Microsoft Edge Security Updates",
            "url": "https://learn.microsoft.com/en-us/deployedge/microsoft-edge-relnotes-security",
            "locator": "September 4,2026 paragraph; November 9,2026 notice"
          },
          {
            "id": "S06",
            "title": "Stable Channel Update for Desktop",
            "url": "https://chromereleases.googleblog.com/2026/09/stable-channel-update-for-desktop_0808145027.html",
            "locator": "Opening release paragraph; CVE-2026-87491 entry and exploitation statement"
          },
          {
            "id": "S07",
            "title": "CVE-2026-87491",
            "url": "https://cveawg.mitre.org/api/cve/CVE-2026-87491",
            "locator": "containers.cna.descriptions and affected"
          },
          {
            "id": "S04",
            "title": "Known Exploited Vulnerabilities catalog",
            "url": "https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json",
            "locator": "vulnerabilities entry matching the named CVE"
          }
        ]
      },
      {
        "id": "chrome-skia-correction",
        "title": "A corrected release note changes the verified fix boundary",
        "kind": "published-vendor-advisory",
        "cves": [
          "CVE-2026-3909"
        ],
        "severity": "High (Google)",
        "affected": {
          "product": "Google Chrome / Skia",
          "reportedRange": "Retrieved CNA description says before 146.0.7680.75; complete affected interval is unresolved alongside the corrected vendor release.",
          "completeHistoryVerified": false
        },
        "fixed": [
          {
            "product": "Chrome Stable",
            "platform": "Windows/macOS/Linux",
            "versions": [
              "146.0.7680.80"
            ],
            "releaseDate": "2026-03-13",
            "correction": "The March 12 .75/.76 note was corrected March 13 to remove the earlier inclusion of this fix."
          }
        ],
        "exploitation": {
          "vendor": "Google reports an exploit in the wild.",
          "kev": {
            "listed": true,
            "dateAdded": "2026-03-13"
          }
        },
        "interpretation": "Use the corrected vendor release for the fix claim. Preserve the CNA discrepancy rather than clearing .75–.79 from an incomplete cutoff.",
        "certainty": {
          "reportedFacts": "high",
          "currentInstalledRisk": "not determined",
          "basis": "Both Google official release notes retrieved through the search index, plus a fresh CVE API record and CISA catalog. The .75/.80 discrepancy remains present."
        },
        "limits": [
          "No claim that every older Skia-containing product shares the same patch boundary.",
          "Correction is visible; unknown affected history stays unknown."
        ],
        "sources": [
          {
            "id": "S08",
            "title": "Stable Channel Update for Desktop, corrected",
            "url": "https://chromereleases.googleblog.com/2026/03/stable-channel-update-for-desktop_12.html",
            "locator": "Updated 2026-03-13 correction above security list"
          },
          {
            "id": "S09",
            "title": "Stable Channel Update for Desktop",
            "url": "https://chromereleases.googleblog.com/2026/03/stable-channel-update-for-desktop_13.html",
            "locator": "Opening release paragraph; sole CVE-2026-3909 entry and exploitation statement"
          },
          {
            "id": "S10",
            "title": "CVE-2026-3909",
            "url": "https://cveawg.mitre.org/api/cve/CVE-2026-3909",
            "locator": "containers.cna.descriptions and affected[0].versions"
          },
          {
            "id": "S04",
            "title": "Known Exploited Vulnerabilities catalog",
            "url": "https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json",
            "locator": "vulnerabilities entry matching the named CVE"
          }
        ]
      },
      {
        "id": "firefox-wasm",
        "title": "Public exploit code does not establish observed attacks",
        "kind": "published-vendor-advisory",
        "cves": [
          "CVE-2026-15718"
        ],
        "severity": "Critical (Mozilla)",
        "affected": {
          "product": "Firefox / Firefox ESR",
          "component": "JavaScript: WebAssembly invalid pointer",
          "reportedRange": "Fix markers verified; complete affected-version history not supplied by the opened advisory/CNA combination.",
          "completeHistoryVerified": false
        },
        "fixed": [
          {
            "product": "Firefox",
            "platform": "As scoped by Mozilla advisory",
            "versions": [
              "152.0.6"
            ],
            "releaseDate": "2026-07-14"
          },
          {
            "product": "Firefox ESR",
            "platform": "As scoped by Mozilla advisory",
            "versions": [
              "140.13"
            ],
            "releaseDate": "2026-07-21"
          }
        ],
        "exploitation": {
          "vendor": "Mozilla acknowledges public exploit code but says no attacks were known at the dated advisories.",
          "kev": {
            "listed": false,
            "qualification": "No match in the retrieved catalog; not proof that exploitation never occurred."
          }
        },
        "interpretation": "Do not turn the publication of exploit code into an observed-attack claim, or the historical lack of known attacks into a permanent assurance.",
        "certainty": {
          "reportedFacts": "high",
          "currentInstalledRisk": "not determined",
          "basis": "Both Mozilla advisories retrieved directly, plus a fresh CVE API record and CISA catalog. Historical public-code and observed-attack statements kept separate."
        },
        "limits": [
          "No exploit code used in this review.",
          "No vulnerable interval invented from branch fix markers.",
          "No local Firefox result inferred."
        ],
        "sources": [
          {
            "id": "S11",
            "title": "MFSA 2026-67: Security Vulnerabilities fixed in Firefox 152.0.6",
            "url": "https://www.mozilla.org/en-US/security/advisories/mfsa2026-67/",
            "locator": "Fixed in; #CVE-2026-15718 description"
          },
          {
            "id": "S12",
            "title": "MFSA 2026-70: Security Vulnerabilities fixed in Firefox ESR 140.13",
            "url": "https://www.mozilla.org/en-US/security/advisories/mfsa2026-70/",
            "locator": "Fixed in; #CVE-2026-15718 description"
          },
          {
            "id": "S13",
            "title": "CVE-2026-15718",
            "url": "https://cveawg.mitre.org/api/cve/CVE-2026-15718",
            "locator": "containers.cna.descriptions and affected"
          },
          {
            "id": "S04",
            "title": "Known Exploited Vulnerabilities catalog",
            "url": "https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json",
            "locator": "vulnerabilities entry matching the named CVE"
          }
        ]
      },
      {
        "id": "safari-webkit",
        "title": "Apple's exploitation statement has a specific iOS scope",
        "kind": "published-vendor-advisory",
        "cves": [
          "CVE-2025-43529",
          "CVE-2025-14174"
        ],
        "severity": "Apple describes code execution/memory corruption; no numeric severity substituted",
        "affected": {
          "product": "Safari / WebKit",
          "reportedRange": "Apple CNA lists Safari before 26.2 for CVE-2025-43529. Other platform branches require their own advisory.",
          "completeHistoryVerified": false
        },
        "fixed": [
          {
            "product": "Safari",
            "platform": "macOS Sonoma and macOS Sequoia",
            "versions": [
              "26.2"
            ],
            "releaseDate": "2025-12-12"
          }
        ],
        "exploitation": {
          "vendor": "Apple reports possible sophisticated targeted exploitation on iOS versions before 26 and links the two CVEs.",
          "kev": {
            "listed": true,
            "entries": [
              {
                "cve": "CVE-2025-43529",
                "dateAdded": "2025-12-15"
              },
              {
                "cve": "CVE-2025-14174",
                "dateAdded": "2025-12-12"
              }
            ]
          }
        },
        "interpretation": "Safari fix applicability and historical iOS exploitation scope are separate facts. A WebKit runtime number is not a Safari patch level.",
        "certainty": {
          "reportedFacts": "high",
          "currentInstalledRisk": "not determined",
          "basis": "Official Apple Safari advisory text retrieved through the search index in English and French localizations, plus fresh CVE API records and CISA catalog. iOS exploitation scope retained."
        },
        "limits": [
          "Do not claim observed attacks on every Mac or all WebKit consumers.",
          "The shared second CVE has a Chrome-centric CNA record; Apple's advisory supplies its platform context."
        ],
        "sources": [
          {
            "id": "S14",
            "title": "About the security content of Safari 26.2",
            "url": "https://support.apple.com/en-us/125892",
            "locator": "Safari 26.2 release heading; WebKit entries for CVE-2025-43529 and CVE-2025-14174",
            "revalidationUrl": "https://support.apple.com/en-hk/125892",
            "revalidationMethod": "Search-indexed official English localization; original US URL returned a rate limit in the web tool."
          },
          {
            "id": "S15",
            "title": "CVE-2025-43529",
            "url": "https://cveawg.mitre.org/api/cve/CVE-2025-43529",
            "locator": "containers.cna.affected product Safari; descriptions"
          },
          {
            "id": "S16",
            "title": "CVE-2025-14174",
            "url": "https://cveawg.mitre.org/api/cve/CVE-2025-14174",
            "locator": "containers.cna.affected and references"
          },
          {
            "id": "S04",
            "title": "Known Exploited Vulnerabilities catalog",
            "url": "https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json",
            "locator": "vulnerabilities entry matching the named CVE"
          }
        ]
      },
      {
        "id": "boundaries",
        "title": "Compatibility checks do not certify browser security",
        "kind": "architecture-and-coverage-boundary",
        "cves": [],
        "severity": null,
        "affected": {
          "product": "Chromium, Gecko, Playwright test engines",
          "reportedRange": "Not applicable",
          "completeHistoryVerified": false
        },
        "fixed": [],
        "exploitation": {
          "vendor": "Not an exploitation claim."
        },
        "interpretation": "Sandbox/site-isolation architecture has boundaries. Playwright's patched engines are not equivalent to every branded browser. A working application test cannot establish patch completeness or absence of vulnerabilities.",
        "certainty": {
          "reportedFacts": "high",
          "currentInstalledRisk": "not determined",
          "basis": "Direct official publications; scope and discrepancies retained."
        },
        "limits": [
          "Verify exact product/platform/channel and current running version separately.",
          "Google's About page checks updates; Relaunch applies an available update.",
          "No browser-security ranking or local inventory is included."
        ],
        "sources": [
          {
            "id": "S17",
            "title": "Site Isolation",
            "url": "https://www.chromium.org/Home/chromium-security/site-isolation/",
            "locator": "Overview; Motivation; Limitations"
          },
          {
            "id": "S18",
            "title": "Firefox Source Docs: Process Model",
            "url": "https://firefox-source-docs.mozilla.org/dom/ipc/process_model.html",
            "locator": "Content Process; Isolated Web Content"
          },
          {
            "id": "S19",
            "title": "Browsers",
            "url": "https://playwright.dev/docs/browsers#webkit",
            "locator": "Google Chrome & Microsoft Edge; Firefox; WebKit"
          },
          {
            "id": "S20",
            "title": "Update Google Chrome",
            "url": "https://support.google.com/chrome/answer/95414?hl=en",
            "locator": "Get a Chrome update when available; Check your current browser version"
          }
        ]
      }
    ],
    "followUpRetrievalLimits": "The five advisory groups were independently rechecked against publisher records, six live CVE API records and CISA catalog 2026.11.10. Google release details and Apple Safari details were also checked through search-indexed official pages, including an official Apple English localization, after some direct web-tool fetches failed. This confirms the historical disclosures and named fix markers; it does not establish current exploitation or installed patch status.",
    "verificationScope": "Targeted revalidation of the five existing advisory groups; not a systematic vulnerability search or a new assessment of architecture documentation."
  },
  "identityRegressionSummary": {
    "originalCoreFixtures": 12,
    "originalBrowserReproductions": 2,
    "afterFixTargetedBrowserCases": 3,
    "passed": 3,
    "unexpectedExternalRequests": 0,
    "pageErrors": 0,
    "marginCoreChecks": 28,
    "converterCoreChecks": 12,
    "limits": "Observed local synthetic workflows. No real inventory import, monetary-loss measurement, third-party scan or browser-engine exploit."
  },
  "startupRegressionSummary": {
    "method": "Fresh profiles on an owner-controlled loopback copy. Hold the application module request, attempt one early sample click or synthetic file selection, release it, and repeat after initialization. Non-loopback and non-GET/HEAD requests blocked.",
    "environments": [
      {
        "browser": "Microsoft Edge",
        "version": "157.0.4341.46"
      },
      {
        "browser": "Google Chrome",
        "version": "156.0.8078.48"
      },
      {
        "browser": "Playwright WebKit on Windows",
        "version": "26.5",
        "qualification": "Patched test build, not branded Safari or a physical iPhone."
      }
    ],
    "baselineCases": 24,
    "baselineLostActions": 21,
    "baselineGatedNegativeControls": 3,
    "afterFixDelayedCases": 24,
    "afterFixControlsDisabledBeforeReady": 24,
    "afterFixReadyActionsWorked": 24,
    "applicationReleases": {
      "/margin-guard/": 6,
      "/supplier-file-converter/": 3,
      "/supplier-quote-comparator/": 2,
      "/order-confirmation-auditor/": 2
    },
    "certainty": "Direct observations support only the named versions and controlled scenarios. Frequency in ordinary traffic, Firefox, extensions and physical mobile devices remain unmeasured."
  },
  "recoveryRegressionSummary": {
    "application": "/order-confirmation-auditor/",
    "release": 3,
    "environments": [
      {
        "browser": "Microsoft Edge",
        "version": "157.0.4341.46"
      },
      {
        "browser": "Playwright WebKit on Windows",
        "version": "26.5",
        "qualification": "Patched test build, not Apple Safari or a physical phone."
      }
    ],
    "method": "Fresh profiles on an owner-controlled loopback copy with fictional CSVs. Complete an annotated review, select a different example, then fail one replacement sample request locally. Compare exact downloaded input hashes and annotations before and after. Keep selector-only and successful replacement controls.",
    "baselineFailedLoadReproductions": 2,
    "independentRootReproductions": 1,
    "afterFixStateCases": 23,
    "passed": 23,
    "originalFilesDeleted": false,
    "externalFaultTesting": false,
    "certainty": "Direct observations of the stated application and scenarios. No actual visitor incidence or customer-loss claim. Historical advisory evidence elsewhere in this report was not expanded by this application check."
  },
  "caseStudies": [
    {
      "url": "https://tenthirtytwo.com/browser-audit/unicode-contact-recovery/",
      "title": "How a split Unicode character cleared visible results",
      "findingIds": [
        "FMP-UNICODE-001"
      ],
      "scope": "Reproduced and repaired owned-application rendering, with independent controls and explicit producer boundaries; no browser-engine claim."
    },
    {
      "url": "https://tenthirtytwo.com/browser-audit/nested-source-metadata/",
      "title": "When nested source metadata interrupted a listing",
      "findingIds": [
        "FMP-METADATA-001"
      ],
      "scope": "Reproduced and repaired owned-application rendering, with independent controls and explicit producer boundaries; no browser-engine claim."
    },
    {
      "url": "https://tenthirtytwo.com/browser-audit/category-catalog-recovery/",
      "title": "When broken suggestions stop a typed search",
      "findingIds": [
        "FMP-CATALOG-001"
      ],
      "scope": "Reproduced and repaired owned-application behavior with explicit controls and limits; no browser-engine claim."
    },
    {
      "url": "https://tenthirtytwo.com/browser-audit/saved-list-recovery/",
      "title": "Recovering a saved list without overwriting it",
      "findingIds": [
        "FMP-RESTORE-001"
      ],
      "scope": "Reproduced and repaired owned-application behavior with explicit controls and limits; no browser-engine claim."
    },
    {
      "url": "https://tenthirtytwo.com/browser-audit/inherited-source-name/",
      "title": "When an inherited source name cleared visible results",
      "findingIds": [
        "FMP-SOURCE-001"
      ],
      "scope": "Reproduced and repaired owned-application behavior with explicit controls and limits; no browser-engine claim."
    },
    {
      "url": "https://tenthirtytwo.com/browser-audit/file-import-review-recovery/",
      "title": "When a failed file load erases the previous review",
      "findingIds": [
        "OCA-RECOVERY-001"
      ],
      "scope": "Expanded explanation of existing verified application findings; no new browser-engine claim."
    },
    {
      "url": "https://tenthirtytwo.com/browser-audit/invisible-sku-csv-validation/",
      "title": "Invisible Unicode identifiers in CSV exports",
      "findingIds": [
        "SFC-IDENTITY-001",
        "MG-IDENTITY-002"
      ],
      "scope": "Expanded explanation of existing verified application findings; no new browser-engine claim."
    },
    {
      "url": "https://tenthirtytwo.com/browser-audit/large-csv-browser-freeze/",
      "title": "Why a small CSV produced 437,704 DOM elements",
      "findingIds": [
        "MG-RESOURCE-001"
      ],
      "scope": "Expanded explanation of existing verified application findings; no new browser-engine claim."
    },
    {
      "url": "https://tenthirtytwo.com/browser-audit/malformed-search-response/",
      "title": "When a malformed response erased good search results",
      "findingIds": [
        "FMP-RESPONSE-001"
      ],
      "scope": "Reproduced and repaired owned-application behavior with explicit controls and limits; no browser-engine claim."
    }
  ],
  "platformControlSummary": {
    "classification": "Bounded positive and negative controls; no browser-engine finding",
    "environments": [
      {
        "browser": "Microsoft Edge",
        "version": "157.0.4341.46"
      },
      {
        "browser": "Playwright WebKit on Windows",
        "version": "26.5",
        "qualification": "Patched test build, not physical Safari or a mobile device."
      }
    ],
    "scenariosPerEngine": 6,
    "total": 12,
    "passed": 12,
    "receipt": "platform-controls-04",
    "method": "Fresh profiles, two loopback origins and fictional fixtures. Non-loopback requests were intercepted with synthetic responses or blocked. Browser sandboxes remained enabled.",
    "controls": [
      "Same-origin DOM allowed; different loopback port denied",
      "Same-origin storage shared; other origin and separate context isolated",
      "Restrictive and permissive sandbox fixtures contrasted",
      "Actual My People noopener link checked locally",
      "Actual application CSP checked locally",
      "Frame-denial configuration mirrored locally"
    ],
    "limits": "Generic platform fixtures are distinct from the actual app link/CSP checks. Local frame-denial mirroring is not live-header verification. No universal safety, account isolation, browser-engine flaw or physical-device compatibility is established. Original WebKit timing-harness failure is retained privately and is not an isolation bypass."
  },
  "deepInputRegressionSummary": {
    "application": "/find-my-people",
    "release": 7,
    "baselineObservations": 22,
    "independentBaselineObservations": 28,
    "repairedObservations": 28,
    "repairedPassed": 28,
    "actualSaveReloadJourneys": 22,
    "sourceRecordPreserved": true,
    "scope": "Two application findings plus passing branch controls and adjacent metadata cases. The independent observations deliberately reproduced defects; they are not repaired passes. Valid Unicode truncation was demonstrated in actual producer functions and a mocked-cache path. No native producer of the malformed nested identifier object was established.",
    "limits": "The producer truncation remains unchanged. Consumer fixes preserve underlying records and ignore unsupported optional attribution. No production occurrence, security compromise, browser-engine flaw, attacker access, customer loss or physical-device result established."
  },
  "externalAssessments": [
    {
      "id": "OUTLOOK-OPINION-001",
      "url": "https://tenthirtytwo.com/browser-audit/outlook-outage-assessment/",
      "title": "Outlook outage: recovery before speculation",
      "classification": "Public-source opinion; no verified fix",
      "confirmedFinding": false,
      "countsAsRepairedFinding": false,
      "source": "https://status.cloud.microsoft/",
      "sourceUpdate": "2026-11-12T19:30:36Z",
      "limits": [
        "Microsoft report, not independent reproduction",
        "Root cause and exploitation unresolved in the reviewed update",
        "Recommendations not tested on Microsoft infrastructure",
        "Publication does not establish Microsoft receipt or endorsement"
      ]
    }
  ],
  "handSignalAudit": {
    "id": "HS-BOUNDARY-001",
    "application": "/office/",
    "applicationRelease": 47,
    "classification": "Targeted security checks and a documented storage boundary; no new exploitable vulnerability reproduced",
    "observationCount": 108,
    "groups": [
      {
        "engine": "Edge",
        "version": "157.0.4341.60",
        "name": "literal HTML and slide text",
        "observations": 12,
        "status": "passed"
      },
      {
        "engine": "Edge",
        "version": "157.0.4341.60",
        "name": "URL protocols, credentials and valid link control",
        "observations": 6,
        "status": "passed"
      },
      {
        "engine": "Edge",
        "version": "157.0.4341.60",
        "name": "CSV formula refusal and ordinary text controls",
        "observations": 13,
        "status": "passed"
      },
      {
        "engine": "Edge",
        "version": "157.0.4341.60",
        "name": "archive boundaries and image type",
        "observations": 5,
        "status": "passed"
      },
      {
        "engine": "Edge",
        "version": "157.0.4341.60",
        "name": "renamed presentation export XML parts",
        "observations": 13,
        "status": "passed"
      },
      {
        "engine": "Edge",
        "version": "157.0.4341.60",
        "name": "invalid workspace files preserve existing document",
        "observations": 3,
        "status": "passed"
      },
      {
        "engine": "Edge",
        "version": "157.0.4341.60",
        "name": "documented same-origin browser storage access",
        "observations": 1,
        "status": "passed"
      },
      {
        "engine": "Edge",
        "version": "157.0.4341.60",
        "name": "fresh profile has no saved copy",
        "observations": 1,
        "status": "passed"
      },
      {
        "engine": "Playwright WebKit",
        "version": "26.5",
        "name": "literal HTML and slide text",
        "observations": 12,
        "status": "passed"
      },
      {
        "engine": "Playwright WebKit",
        "version": "26.5",
        "name": "URL protocols, credentials and valid link control",
        "observations": 6,
        "status": "passed"
      },
      {
        "engine": "Playwright WebKit",
        "version": "26.5",
        "name": "CSV formula refusal and ordinary text controls",
        "observations": 13,
        "status": "passed"
      },
      {
        "engine": "Playwright WebKit",
        "version": "26.5",
        "name": "archive boundaries and image type",
        "observations": 5,
        "status": "passed"
      },
      {
        "engine": "Playwright WebKit",
        "version": "26.5",
        "name": "renamed presentation export XML parts",
        "observations": 13,
        "status": "passed"
      },
      {
        "engine": "Playwright WebKit",
        "version": "26.5",
        "name": "invalid workspace files preserve existing document",
        "observations": 3,
        "status": "passed"
      },
      {
        "engine": "Playwright WebKit",
        "version": "26.5",
        "name": "documented same-origin browser storage access",
        "observations": 1,
        "status": "passed"
      },
      {
        "engine": "Playwright WebKit",
        "version": "26.5",
        "name": "fresh profile has no saved copy",
        "observations": 1,
        "status": "passed"
      }
    ],
    "externalRequestsObserved": 0,
    "nonGetHeadRequestsObserved": 0,
    "method": "Actual local application modules, parsed generated HTML and XML, file-opening UI, explicit synthetic storage save, same-origin path read, and a separate fresh-profile control. Owned loopback only.",
    "preconditions": "A visitor opens a selected file, exports typed content, or explicitly saves a copy in this browser profile. The storage observation requires code with same-origin access.",
    "storageConclusion": "Same-origin paths can read the deliberately saved device copy. This matches the existing notice; it is not a new remote exploit or evidence of stolen records.",
    "limits": "Synthetic owned loopback tests. No server authorization, dependency, browser-engine, extension, native Office or physical-device security certification. Same-origin storage access is a documented design boundary, not a newly discovered remote exploit.",
    "additionalLimits": [
      "Markup checks inspect parsed generated HTML; they are not a complete active exploit campaign.",
      "CSV controls use the application parser; native Excel and LibreOffice were not tested.",
      "26 of the 108 observations are XML-part parsing checks, not attack cases.",
      "Both browser runs use the same tester and harness; no independent external replication."
    ],
    "sources": [
      {
        "title": "OWASP CSV Injection",
        "url": "https://owasp.org/www-community/attacks/CSV_Injection"
      },
      {
        "title": "MDN IndexedDB API",
        "url": "https://developer.mozilla.org/en-US/docs/Web/API/IndexedDB_API"
      },
      {
        "title": "OWASP XSS Prevention",
        "url": "https://cheatsheetseries.owasp.org/cheatsheets/Cross_Site_Scripting_Prevention_Cheat_Sheet.html"
      }
    ]
  },
  "programBusters": {
    "url": "https://www.tenthirtytwo.com/program-busters/",
    "scope": "Application workflows, file handling, open privacy boundaries and selected upstream advisories."
  }
}
