Manifest position 148 · CrUX rank bucket 1000
https://tw.stock.yahoo.com
Coverage complete
Coverage complete. Public signed-out representative templates only; authenticated, payment and external-host journeys were excluded as recorded.
All 58 atomic check outcomes
| Principle / check | Status | Confidence | Evidence or reason |
|---|---|---|---|
respect-user-preferencesrespects-color-scheme | issues | high | The dark-condition screenshot remains the same white interface as the default capture, and the live CSS probe found no prefers-color-scheme, light-dark(), or color-scheme implementation. F001 medium: The site ignores the user’s dark-color preference. |
respect-user-preferencesrespects-reduced-motion | pass | medium | Under prefers-reduced-motion: reduce, matchMedia returned true, getAnimations() returned no active animation, and scroll-behavior was auto. |
respect-user-preferencesrespects-contrast | issues | high | The prefers-contrast: more capture does not materially change controls or text, the CSS probe found no prefers-contrast or forced-colors rules, and Lighthouse reports color-contrast failures. F002 medium: High-contrast preferences receive no dedicated treatment. |
implement-natural-interactionsview-transitions | issues | high | The cross-template CSS probe found no view-transition rules or API styling on home, quote, listing, news, or screener surfaces. F003 low: Route and state changes have no View Transition treatment. |
implement-natural-interactionsscroll-driven-animations | pass | medium | CSS inspection across all five templates found no animation-timeline/scroll-timeline and visual captures showed no scroll-linked parallax or reveal behavior, so no main-thread scroll animation anti-pattern was present. |
implement-natural-interactionsphysical-gestures | not-applicable | high | No custom swipe, pull, drag, carousel or other gesture-driven control was present on the five representative signed-out surfaces. |
provide-guided-navigationscroll-state-aware-chrome | issues | high | The 6,617px desktop landing capture has static chrome and no progress cue; CSS inspection found no scroll-state query or scroll-timeline implementation. F004 low: The long landing page gives no scroll-position feedback. |
provide-guided-navigationanchored-positioning | not-applicable | high | No tooltip, transient popover or edge-positioned menu was exposed in the representative signed-out paths, so anchor positioning had no applicable target. |
provide-guided-navigationdirects-attention | pass | medium | Screenshots show selected global and local navigation states (purple active tabs), breadcrumbs on screener, and descriptive route headings that preserve location across templates. |
maximize-content-reduce-noiseno-intrusive-interruptions | pass | medium | Initial screenshots across five paths show no full-viewport modal, consent wall or interstitial obscuring the core content; banners remain in document flow. |
maximize-content-reduce-noisesemantic-dismissible-primitives | not-applicable | high | No modal, dialog, disclosure or transient overlay was present during load or the signed-out screener state. |
maximize-content-reduce-noisereduced-chrome | issues | high | The home and screener captures show multiple large banners, a right-rail ad, native-ad rows interleaved with news, and repeated navigation/search bars before core content. F005 high: Advertising and auxiliary chrome compete with financial content. |
adapt-to-the-form-factorresponsive-no-horizontal-scroll | issues | high | All five 360x800 layout captures report clientWidth 1,020 and hasViewportMeta false; screenshots are desktop layouts shrunk to roughly 35%, making text and controls extremely small. F006 critical: The requested mobile viewport renders a scaled 1,020px desktop canvas. |
adapt-to-the-form-factorcomponent-level-responsiveness | issues | high | CSS probes across all five templates found no @container rules; dense cards and tables retain desktop arrangements in the scaled mobile canvas. F007 medium: Reusable components do not adapt to their available container. |
adapt-to-the-form-factorinput-modality-aware | issues | high | The custom probe found 50 visible controls under 24px on the home page, and Lighthouse failed target-size; this is amplified by the 0.35 mobile scale. F008 high: Many interactive targets are too small for touch. |
support-core-task-successclear-purpose-and-primary-action | pass | medium | Home exposes market conditions immediately, quote identifies 台積電 with live quote content, listing labels its market groups, news identifies 財經新聞, and screener exposes predefined strategies and login boundary. |
support-core-task-successprimary-flow-completion | pass | medium | The public stock-research flow completed from home to the TSMC quote and related market/news surfaces without an account wall; the screener provides usable predefined strategy results while clearly gating only custom strategy persistence. |
support-core-task-successclear-system-state-and-recovery | pass | medium | The screener explicitly states 尚未登入Yahoo帳號 and offers a login action while preserving predefined strategies; the tested invalid URL returns an actual HTTP 404 rather than a false success. |
be-fast-and-stablegood-core-web-vitals | issues | high | Lighthouse measured home LCP 8.5s, TBT 610ms and performance 0.44; screener measured LCP 4.6s, CLS 0.289, TBT 580ms and performance 0.38. F009 high: Primary templates miss the performance and stability baseline. |
be-fast-and-stablevisual-stability | issues | high | Lighthouse measured CLS 0.289 on screener; the image inventory found 25 of 30 home images without width/height attributes. F010 medium: The screener and image markup allow visible layout movement. |
be-fast-and-stableefficient-main-thread | issues | high | The trace recorded three long tasks, longest 196ms and TBT 184ms under a warm local capture; Lighthouse under throttling measured 610ms TBT and 27.7s to interactive. F011 high: JavaScript blocks responsiveness during load. |
be-fast-and-stableefficient-resource-delivery | issues | high | HAR captured 222 requests and 3.57MB transferred, including 2.50MB script and 3.42MB classified third-party; it also identified eight first-party parser-inserted VeryHigh-priority stylesheets before third-party styles. F012 high: The critical load is dominated by scripts and third parties. |
be-fast-and-stabletrim-unused-and-duplicate-code | issues | high | HAR largest scripts include Taboola 217KB, Google Publisher 200KB, Prebid 184KB and GTM 166KB; Lighthouse also reports missing source maps for large first-party JavaScript. F013 high: Large advertising and analytics bundles dominate shipped code. |
be-inclusivenames-roles-labels | issues | high | Lighthouse fails link-name and label-content-name-mismatch; the image inventory reports 20 of 30 images without alt, and 85 of 294 home links have no discernible text in the DOM probe. F014 high: Important controls and content lack accessible names. |
be-inclusivesufficient-contrast | issues | high | Lighthouse’s color-contrast audit fails on both home and screener, corroborated by the unchanged high-contrast preference capture. F015 medium: Text and essential UI fail minimum contrast. |
be-inclusivestructure-and-focus | issues | high | All five DOM probes report zero main landmarks; Lighthouse fails landmark-one-main and link-name, while quote has two h1 elements and screener uses another h1 for the login state. F016 high: Pages have no main landmark and contain noisy heading/link structure. |
be-inclusivelegible-text | pass | medium | Desktop captures show stable, unclipped Traditional Chinese text with consistent line wrapping and clear numeric alignment; legibility problems are isolated to the separately reported mobile scaling failure. |
be-inclusivezoom-reflow-targets-and-media | issues | high | Lighthouse fails meta-viewport because user-scalable=no or maximum-scale below 5 is present; CDP mobile layouts report hasViewportMeta false and a 1,020px layout scaled to 360px. F017 high: Viewport policy prevents normal zoom and reflow. |
follow-best-practicesno-console-errors | issues | high | Lighthouse fails errors-in-console on both home and screener; the HAR also records failed consent/cookie-sync requests and aborted analytics/ad requests. F018 medium: The page emits browser errors during normal load. |
follow-best-practicessound-document-and-assets | issues | high | The image audit found 25/30 images without intrinsic dimensions, four oversized images and six legacy-format assets; Lighthouse reports low-resolution image delivery. F019 medium: Asset markup is not consistently sound. |
follow-best-practicesbrowser-platform-hygiene | issues | high | Lighthouse reports deprecated API usage, missing source maps, inspector issues and a BFCache failure on both tested templates. F020 medium: The page has several platform-hygiene failures. |
be-discoverabletitle-and-description | pass | medium | All five DOM probes found unique descriptive titles and meta descriptions specific to home, quote, market listing, news and screener. |
be-discoverablecrawlable-and-mobile-friendly | issues | high | Lighthouse fails crawlable-anchors and meta-viewport; the probe counts 85 empty-name links on home and 77 on screener. F021 high: Some navigation is unnamed or not crawlable, and mobile metadata is harmful. |
be-discoverablecanonical-and-indexing-signals | pass | medium | All representative pages returned HTTP 200 and expose self-canonical URLs; robots.txt references working sitemap indexes and no accidental noindex was observed. |
be-discoverablestructured-and-shareable-metadata | pass | medium | All five probes found six Open Graph fields; home/class/screener also include JSON-LD, providing share metadata appropriate to these listing surfaces. |
be-private-and-securesecure-transport-and-headers | issues | high | HTTPS, HSTS and nosniff are present, but Content-Security-Policy contains only frame-ancestors and does not constrain script-src, style-src, object-src or base-uri. F022 medium: The CSP provides framing control but little script/style XSS defense. |
be-private-and-securedata-minimisation-and-third-parties | issues | high | Tracker evidence found 25 third-party origins and 11 known tracker domains; HAR classified 217/222 requests and 3.42MB/3.57MB as third-party, including Google, Taboola and PubMatic systems. F023 high: The default load has a very large tracking and third-party footprint. |
be-private-and-securein-context-permissions-and-modern-auth | pass | medium | No permission element or prompt appeared on load across representative paths; authentication is delegated to login.yahoo.com and no password form is hosted on the audited origin. |
be-private-and-securedefensive-browser-policies | issues | high | The header audit found no Permissions-Policy and only a legacy no-referrer-when-downgrade Referrer-Policy; framing is covered by CSP frame-ancestors and HSTS is present. F024 medium: Browser defense headers are incomplete. |
be-resilientprogressive-enhancement | issues | high | Discoverability comparisons found only 25% raw/rendered word coverage on home, 16% on the TSMC quote and 31% on news, although titles and h1 survive. F025 high: Core dashboard, quote and news content is largely absent without JavaScript. |
be-resilientresilient-runtime-behaviour | pass | medium | Five rendered paths completed without blank shells or clipped modal/menu states; class and screener preserve 88–100% raw content and the public route transitions resolve to stable documents. |
be-resilientoffline-and-installable | not-applicable | high | Live market prices and screening are intrinsically online; no offline/installable intent was declared, so the contextual PWA check is not applicable. |
be-resilientnetwork-and-http-failure-states | issues | high | A missing stock route correctly returns HTTP 404, but its body only redirects with JavaScript or noscript meta refresh to the general Yahoo homepage error surface. F026 low: Unknown routes depend on script or meta refresh and lose site context. |
be-internationalisedlang-dir-and-logical-properties | issues | high | All probes report lang=zh-Hant-TW, but no logical properties were detected and dozens of physical left/right declarations were present on home, quote and news. F027 low: The document language is correct but layout CSS is not writing-mode resilient. |
be-internationalisedlocale-aware-data | pass | medium | Rendered pages use zh-Hant-TW labels, localized relative durations, comma-separated numbers and market/currency notation consistently across home, quote and listings. |
be-internationalisedtime-zone-correctness | pass | medium | Visible market update times and relative news durations are rendered consistently in the zh-Hant-TW experience without ambiguous raw timestamp strings. |
be-trustworthyno-dark-patterns | issues | high | The home feed screenshot includes rows such as “Search For Booking Software” and “Find Results…” styled like news cards with only a small Ad label, alongside multiple high-prominence banners. F028 high: Native-looking advertisements are interleaved with editorial results. |
be-trustworthyhumane-error-handling | not-applicable | high | No authored data-entry form with validation was available in the representative signed-out scope; stock search is suggestion-driven and has no required-field submission. |
be-trustworthytrustworthy-input-assistance | not-applicable | high | The only first-party text input is stock-symbol search; address, payment, sign-in and sign-up fields are hosted outside the audited origin. |
be-trustworthysafe-commercial-and-account-flows | not-applicable | high | VIP purchase, authentication and account management require credentials/payment or move to external Yahoo services and were explicitly excluded from this public signed-out audit. |
be-sustainableoptimised-assets | issues | high | The image audit reports 25 images without dimensions, four oversized, six legacy-format, two without srcset and no images recognized as modern format. F029 medium: Image delivery wastes bytes and lacks intrinsic metadata. |
be-sustainableno-wasteful-work | issues | high | The load makes 222 requests, executes 61 scripts, and records repeated analytics/ad calls plus long tasks after content begins rendering. F030 medium: Non-core background and advertising work is substantial. |
be-sustainablethird-party-and-media-budget | issues | high | HAR attributes 3.42MB of 3.57MB and 217 of 222 requests to third-party origins; the largest non-core scripts include Taboola, Google Publisher, Prebid and GTM. F031 high: Third-party cost is disproportionate to the core finance experience. |
be-agent-readystructured-agent-capabilities | issues | high | robots.txt disallows anthropic-ai, ClaudeBot, GPTBot, ChatGPT-User, Google-Extended, PerplexityBot and many peers; probes found no WebMCP/agentic capability, while no-JS quote coverage is only 16%. F032 medium: The public finance site is deliberately unavailable to major AI agents and exposes no structured agent tools. |
be-agent-readyon-device-inference | not-applicable | high | No summarisation or generative interaction is offered in the current product surface, so an on-device inference implementation has no defined user task to satisfy. |
be-memory-efficientno-leak-under-repeated-interaction | issues | medium | Across baseline and post snapshots, heap self-size rose from 53.5MB to 62.9MB and snapshot nodes from 881,921 to 1,036,618; same-session performance.memory rose from 54.8MB to 69.6MB and DOM nodes from 2,177 to 2,863. Lazy feed loading may explain part of this, so leak confidence is medium. F033 medium: Repeated search/scroll interaction retains substantial additional memory. |
be-memory-efficientbounded-footprint | pass | medium | Baseline heap self-size was 53.5MB with 2,177 live DOM elements in a dense market/news dashboard; this single-state footprint is substantial but proportionate to the rendered content and below the separate interaction-growth concern. |
be-memory-efficientno-detached-dom-or-unbounded-listeners | pass | medium | Neither heap summary exposed a Detached* constructor among the largest retained constructors; the evidence did not show a direct detached-DOM population, with interaction growth recorded separately. |
Provenance
Canonical report: results/atomic/reports/0148-tw_stock_yahoo_com.json
Report SHA-256: 147422476768f80ffad16f0293826d2b18b49352225d89ca9e55bd3ed55309b6
Local retained report: runs/2026-07-17T17-27-24-856Z/atomic-reports/tw_stock_yahoo_com/2026-07-18T18-34-22-365Z/report.json
Local evidence root: runs/2026-07-17T17-27-24-856Z/atomic-reports/tw_stock_yahoo_com/2026-07-18T18-34-22-365Z
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.