Manifest position 963 · CrUX rank bucket 1000
https://www.polybuzz.ai
Coverage complete
All 58 catalog checks received conclusive outcomes. Intermittent EdgeOne 567 blocks were judged as resilience evidence rather than an audit blocker because other browser and raw HTTP runs succeeded.
Attempts
2 / 3
Judged checks
58 / 58
Blocked
0
Not run
0
This report has no published overall score. Blocked and not-run checks are not passes. A coverage-complete report means every check has a judged outcome; it does not mean every check passed.
All 58 atomic check outcomes
| Principle / check | Status | Confidence | Evidence or reason |
|---|---|---|---|
respect-user-preferencesrespects-color-scheme | issues | high | The normal desktop capture is dark, while raw server HTML contains no prefers-color-scheme or color-scheme handling. F01 medium: The site hard-codes a dark presentation instead of adapting to the user’s colour-scheme preference. |
respect-user-preferencesrespects-reduced-motion | issues | high | Raw server HTML and inline styles contain no prefers-reduced-motion rule; the reduced-motion capture was additionally blocked by the edge WAF, so static authored output is the direct signal. F02 medium: Reduced-motion preferences are not implemented. |
respect-user-preferencesrespects-contrast | issues | high | Lighthouse reports a failing colour-contrast node (.fc-faq-label), and raw output contains no prefers-contrast or forced-colors handling. F03 medium: Some text fails minimum contrast and there is no high-contrast adaptation. |
implement-natural-interactionsview-transitions | issues | high | Raw rendered HTML contains no view-transition declarations, while this SPA exposes route changes between discovery, generator, chat and membership surfaces. F04 low: Route and state changes do not use View Transitions. |
implement-natural-interactionsscroll-driven-animations | pass | medium | No parallax, scrollytelling or scroll-linked reveal was observed on representative pages, so no main-thread scroll-animation anti-pattern was present. |
implement-natural-interactionsphysical-gestures | pass | medium | The mobile feed uses native vertical scrolling, has no horizontal overflow, and no custom pointer-driven gesture surface was observed. |
provide-guided-navigationscroll-state-aware-chrome | pass | medium | Persistent desktop side navigation and mobile bottom navigation remain position-aware and available without obscuring the feed. |
provide-guided-navigationanchored-positioning | pass | medium | The observed login promotion and consent surfaces remained within viewport bounds at desktop and mobile sizes; no drifting anchored tooltip/menu was observed. |
provide-guided-navigationdirects-attention | pass | medium | Selected navigation, Back on chat, headings and card hierarchy visibly communicate location and next actions behind the consent layer. |
maximize-content-reduce-noiseno-intrusive-interruptions | issues | high | Desktop and mobile screenshots show a full-screen dimmer and consent modal before any interaction; on mobile it occupies almost the full viewport. F05 high: A consent wall obscures the entire experience on every tested path. |
maximize-content-reduce-noisesemantic-dismissible-primitives | issues | high | Raw output contains no dialog or popover primitive; the chat screenshot shows a blank white modal panel over the conversation. F06 medium: Overlays are implemented as custom/non-semantic surfaces and can render blank. |
maximize-content-reduce-noisereduced-chrome | pass | medium | Outside the consent interruption, the discovery surface dedicates most space to character cards and collapses desktop chrome into a compact mobile bar. |
adapt-to-the-form-factorresponsive-no-horizontal-scroll | pass | medium | Layout reports scrollWidth 360, innerWidth 360 and horizontalOverflowPx 0 with a viewport meta present. |
adapt-to-the-form-factorcomponent-level-responsiveness | issues | high | The SSR output contains no @container or container-type despite card components being reused in desktop grids and mobile feeds. F07 low: Reusable cards and navigation rely on viewport styling rather than container responsiveness. |
adapt-to-the-form-factorinput-modality-aware | issues | high | Lighthouse found four links without discernible names in the mobile footer navigation, preventing reliable keyboard and assistive interaction. F08 medium: Mobile icon navigation has unnamed interactive targets. |
support-core-task-successclear-purpose-and-primary-action | pass | medium | The server-rendered H1 explicitly describes free AI character chat and visible cards/search communicate the primary discovery action. |
support-core-task-successprimary-flow-completion | issues | high | All four representative path screenshots are covered by the consent surface; the chat path additionally shows a blank modal. F09 critical: The consent wall prevents the primary chat, image-generation and membership journeys from starting. |
support-core-task-successclear-system-state-and-recovery | issues | high | The discoverability browser capture intermittently receives EdgeOne “Restricted Access 567” with only an owner-oriented instruction, and the chat overlay is blank. F10 high: Blocked and failed states do not provide useful recovery. |
be-fast-and-stablegood-core-web-vitals | issues | high | Lighthouse measured LCP 20.8 s, FCP 3.9 s, TBT 700 ms and a performance score of 0.40; a separate trace measured LCP 2.25 s, showing high run-to-run variability. F11 high: Mobile loading performance is poor. |
be-fast-and-stablevisual-stability | pass | medium | Lighthouse CLS is 0.093 (good range), while the focused mobile layout run observed CLS 0 and no shifts. |
be-fast-and-stableefficient-main-thread | issues | high | Lighthouse measured 700 ms TBT; 71 scripts were transferred in the HAR. F12 high: Initial JavaScript work blocks the main thread. |
be-fast-and-stableefficient-resource-delivery | issues | high | HAR recorded 173 requests and 2.23 MB, including ten parser-inserted high-priority stylesheets; Lighthouse reports a 6.1 s dependency chain and about 500 ms render-blocking savings. F13 high: The critical network path is long and cache/render delivery is inefficient. |
be-fast-and-stabletrim-unused-and-duplicate-code | issues | high | Lighthouse estimates 572 KiB unused JavaScript and 42 KiB unused CSS, including advertising, tag-manager and app bundles. F14 high: Large amounts of unused JavaScript and CSS ship on first load. |
be-inclusivenames-roles-labels | issues | high | Lighthouse accessibility reports four unnamed mobile footer links; the images primitive also reports eight images without alt text. F15 high: Several interactive links lack accessible names. |
be-inclusivesufficient-contrast | issues | high | Lighthouse identifies .fc-faq-label as failing the colour-contrast audit. F16 medium: Consent/help text has insufficient colour contrast. |
be-inclusivestructure-and-focus | pass | medium | Heading order, aria-allowed attributes and aria-hidden focus checks pass; failures are separately recorded for unnamed links. |
be-inclusivelegible-text | pass | medium | Primary headings, card labels and consent copy use readable sizing and wrapping at both tested viewports, apart from the separately recorded contrast failure. |
be-inclusivezoom-reflow-targets-and-media | issues | high | The page ships maximum-scale=1 and user-scalable=no; Lighthouse fails the meta-viewport accessibility audit. F17 high: The viewport explicitly prevents user zoom. |
follow-best-practicesno-console-errors | pass | medium | Lighthouse reports no browser console errors or uncaught exceptions. |
follow-best-practicessound-document-and-assets | pass | medium | The document has HTML5 doctype and UTF-8; Lighthouse passes image aspect/resolution checks and 48/55 observed images use modern formats. |
follow-best-practicesbrowser-platform-hygiene | issues | high | Lighthouse reports one BFCache restoration failure even though deprecation and inspector audits pass. F18 medium: The page is not eligible for back/forward cache. |
be-discoverabletitle-and-description | pass | medium | The home response contains a descriptive unique title and meta description aligned with AI character chat. |
be-discoverablecrawlable-and-mobile-friendly | issues | high | Lighthouse reports an anchor.delete-account without href and fails the restrictive viewport audit. F19 medium: One account-management action is not a crawlable link and the viewport is not zoom-friendly. |
be-discoverablecanonical-and-indexing-signals | issues | high | GET /sitemap.xml returned HTTP 200 but HTML whose title/canonical/hreflang all describe /404 rather than XML. F20 high: The advertised sitemap endpoint serves a soft 404 document. |
be-discoverablestructured-and-shareable-metadata | pass | medium | The home response contains five Open Graph properties and one JSON-LD block describing visible content. |
be-private-and-securesecure-transport-and-headers | issues | high | Headers evidence shows no CSP or X-Content-Type-Options; all 12 observed cookies lack Secure and most use SameSite=None. F21 high: Security headers and cookie attributes are incomplete. |
be-private-and-securedata-minimisation-and-third-parties | issues | high | HAR records 170/173 requests and 2.19 MB as non-main-origin, including Google Ads/Analytics and Facebook; trackers evidence finds four known trackers while the consent wall is still on screen. F22 critical: Tracking executes before an informed consent choice. |
be-private-and-securein-context-permissions-and-modern-auth | issues | high | The public SSR output has a Log In route but no WebAuthn, PublicKeyCredential or navigator.credentials integration; no permission prompt fired on load. F23 medium: No phishing-resistant authentication capability is exposed. |
be-private-and-securedefensive-browser-policies | issues | high | Headers evidence shows no CSP, Referrer-Policy, Permissions-Policy or X-Content-Type-Options; HSTS and X-Frame-Options are present. F24 high: Browser-enforced defensive policies are missing. |
be-resilientprogressive-enhancement | pass | medium | The 145 KB server response contains about 800 content tokens, an H1 and real character links, so core discovery content is not an empty JS shell. |
be-resilientresilient-runtime-behaviour | issues | high | The chat screenshot shows a large blank white overlay with no label, loading status, close affordance or recovery copy. F25 high: The chat route can leave an empty modal covering the application. |
be-resilientoffline-and-installable | issues | high | HAR loads /manifest.webmanifest, while SSR output contains no service-worker registration and WAF/network failures replace the app with a dead-end page. F26 medium: The app advertises a manifest but has no observable offline worker/fallback. |
be-resilientnetwork-and-http-failure-states | issues | high | The rendered discoverability capture shows Restricted Access 567 and tells the user only to contact the owner; no retry, home or support action is present. F27 high: Network/security failures produce a dead end rather than a recoverable app state. |
be-internationalisedlang-dir-and-logical-properties | pass | medium | The document declares lang=en and exposes 16 hreflang alternates; no incorrect forced RTL direction was present. |
be-internationalisedlocale-aware-data | pass | medium | Requests include localLanguage=en-GB and localTimezone=Europe/London, and the site exposes locale variants rather than one hard-coded locale. |
be-internationalisedtime-zone-correctness | pass | medium | User detail requests explicitly transmit the IANA time zone Europe/London, avoiding an ambiguous numeric offset. |
be-trustworthyno-dark-patterns | issues | high | The first layer offers Consent and Manage options but no equally prominent Reject all; HAR simultaneously records Google/Facebook advertising and analytics requests before choice. F28 critical: Consent is not a balanced, informed choice and trackers run before it. |
be-trustworthyhumane-error-handling | issues | high | The chat screenshot contains an empty white panel over dimmed content with no status text or actionable control. F29 high: The blank chat modal gives no error explanation or recovery. |
be-trustworthytrustworthy-input-assistance | pass | medium | The public entry page exposes search rather than address/payment data collection; no paste prevention or obstructive input behaviour was reported by Lighthouse. |
be-trustworthysafe-commercial-and-account-flows | issues | high | The membership screenshot is an otherwise blank black page beneath the consent wall, so pricing, commitment and cancellation information are unavailable. F30 high: Consent blocks pricing/account evaluation and the membership path renders no product context behind it. |
be-sustainableoptimised-assets | issues | high | Images evidence finds 10 oversized images, 42 without srcset and 7 legacy-format assets; 55 images lack width/height attributes even where CSS may reserve space. F31 medium: Image delivery is not consistently responsive or right-sized. |
be-sustainableno-wasteful-work | issues | high | HAR records 71 scripts and 170 non-main-origin requests; Lighthouse estimates 572 KiB unused JavaScript. F32 high: The page performs substantial non-essential work before consent. |
be-sustainablethird-party-and-media-budget | issues | high | HAR attributes 2.19 MB of 2.23 MB transferred to non-main origins, with GTM, Facebook and AdSense among the largest scripts. F33 high: Third-party code dominates the page budget. |
be-agent-readystructured-agent-capabilities | issues | high | SSR output contains no WebMCP or agentic form/tool declarations despite chat and image generation being suitable structured tasks. F34 low: The AI service exposes no structured browser-agent capability. |
be-agent-readyon-device-inference | issues | high | SSR output contains no built-in LanguageModel or Summarizer integration; all product intelligence is server-oriented. F35 low: No on-device inference path is offered for lightweight AI tasks. |
be-memory-efficientno-leak-under-repeated-interaction | pass | medium | Node count changed 30,610 to 30,752 (+0.46%) and self size 1,024,113 to 1,031,089 bytes (+0.68%), with no unbounded growth signal. |
be-memory-efficientbounded-footprint | pass | medium | The tested state retained 30,610 nodes and about 1.02 MB self size, proportionate to the rendered consent/discovery surface. |
be-memory-efficientno-detached-dom-or-unbounded-listeners | pass | medium | After ten down/up scroll cycles the heap increased by only 142 nodes and about 7 KB; no accumulating population was evident at this test resolution. |
Provenance
Canonical report: results/atomic/reports/0963-www_polybuzz_ai.json
Report SHA-256: 9780f6867a47d8f8427c3add5c4ea4521471c218cd4a0edca3b2f7f15ef30935
Local retained report: runs/2026-07-17T17-27-24-856Z/atomic-reports/www_polybuzz_ai/2026-07-27T06-48-17-689Z/report.json
Local evidence root: runs/2026-07-17T17-27-24-856Z/atomic-reports/www_polybuzz_ai/2026-07-27T06-48-17-689Z
Catalog SHA-256: sha256:78ccfdb2d483f4c57d9dafed80fd86c6265585a56457c8dcfddc254b80fb44d7
Raw screenshots, HARs, traces, heaps, and other browser artifacts are retained at the local evidence root and intentionally are not committed. Artifact paths in the canonical report are relative to that root.