The new media kit brings official creator platforms, available audience metrics, selected recent content, and collaboration contact together. The press page remains the source for approved bios, media downloads, attribution, and source references.
Design and delivery
The page uses a bounded, responsive dark layout with purple accents, existing avatar artwork, readable metric cards, platform jump links, and a print-friendly layout. The approved quick biography is reused without changing the creator’s name or claims. Navigation links connect the media kit with Press, Projects, Work, and Contact.
Primary content and the latest saved numbers are rendered as HTML at build time. Collection runs separately from page requests on the hourly social-feed workflow. A small first-party script checks for newer published JSON every five minutes while the page is visible. It does not embed social widgets, load chart libraries, or call platform APIs from visitors’ browsers.
What the numbers mean
Every metric has an observation date. Unknown values display a dash, not zero. Failed or partial requests preserve the last successful values with their original dates and mark them as older snapshots. The audience sum includes only successfully refreshed follower/subscriber counts observed within six hours; it is not a count of unique people.
Public Bluesky and GitHub metrics are collected without private credentials. Instagram and TikTok collectors reuse the existing encrypted social-feed authorization after its scheduled refresh. YouTube and X collectors activate only when their API credentials are configured. Twitch stream snapshots use the existing app connection; follower totals require a user token. Other official profiles remain visible with explicit connection limits. No individual followers or private account records are published.
Approved aggregate-demographics follow-up
Sammy approved publishing aggregate age, gender, and country insights when an authorized platform provides them, and confirmed the existing Facebook profile as the only Facebook destination. The Instagram collector now requests those three demographic breakdowns with its last-30-days filter. Accessible native meter charts show only recognized returned groups; unavailable or privacy-restricted data stays unavailable. Group percentages are not assumed to represent every follower. No city-level or individual data is collected.
Following counts, Instagram sample-like totals, and GitHub public-repository star totals extend the platform metrics where supported. Stars are labeled separately from likes. TikTok video sample metrics can establish account identity through exact official share URLs when profile scope is absent; account statistics still require their separate permission. The first authenticated collection confirmed Instagram followers, posts, average likes and comments, plus Twitch live-viewer snapshots. Missing access on another platform is not replaced with these numbers or the supplied screenshots.
Recent-post averages are labeled as lifetime counters on a sample of up to 12 posts, not 30-day reach or engagement. The old screenshot totals and demographic charts are not treated as current evidence. Community membership is not counted as creator followers.
Collection record
Creator breakdowns, exported metrics and history displays
The media kit now includes a creator-role breakdown using the approved biography, identity and attribution references, and a collaboration brief checklist. Asset permission guidance continues to point to the press page. No rates, sponsorship results, testimonials or availability guarantees have been invented.
Facebook, Kick, LinkedIn and Reddit now have explicit metric panels ready for owner exports, alongside the seven existing collector platforms. Per-profile reporting coverage distinguishes missing counters from audience quality. Reddit karma remains separate from followers and is not counted in audience totals; audience coverage uses only platforms with a follower/subscriber field. Reporting-period views, reach, impressions, interactions, watch time, clicks, shares and saves can be shown when a dated, owner-approved export supplies them.
Social Poller now parses CSV or JSON exports locally using explicit column mapping. Raw exports and unselected private fields are not uploaded. A separate D1 database records actual aggregate observations; repeated publications are deduplicated, and graphs keep different sources and measurement scopes apart. There is no estimated history or promise that a platform exposes every metric.
The media kit offers history graphs, rotating cards and tickers on demand. Their isolated frames are created only after a visitor requests a display, so no new widget is loaded during the initial page render. Playback is opt-in and honors reduced motion. The CSP permits frames from the exact Social Poller origin only; the existing static cards, canonical indexing policy and robots file are retained.
The expanded release passed a 26-page browser sweep, 394 HTTP/cache checks and twelve focused media-kit behavior checks. Its live mobile Lighthouse sample scored 100 in all four categories, with 1.48-second LCP, 54 ms TBT and zero layout shift. These are measured laboratory results, not guarantees. The hourly publisher successfully stored dated observations in Social Poller; its separate YouTube RSS refresh returned 404 and retained the previous cached posts. The final eleven-panel follow-up again passed twelve focused browser checks; its Lighthouse sample scored 99 Performance and 100 in the other categories, with 2.01-second LCP, 1 ms TBT and zero layout shift. Live robots.txt remains byte-identical to the committed file, and the page keeps index/follow headers.
Separate Social Poller service
Social Poller adds a public exact-profile lookup API and embeddable statistics cards in a separate repository and Cloudflare Worker. GitHub and Bluesky use public APIs. Other available creator metrics arrive from the existing hourly, authenticated collector as sanitized aggregates; platform credentials stay with that collector. Protected JSON imports support explicitly registered owner exports. No social-page scraping or access-control circumvention is used.
The media kit now checks a same-origin service relay after load. If the secondary Worker is unavailable or returns invalid data, the relay serves the original saved JSON. The primary page, counts, biography and links remain build-rendered HTML. The separate utility API uses noindex; canonical creator pages retain their existing index/follow policy and unchanged robots.txt authority.
Public lookups, authorized snapshots and owner exports keep separate cache keys, source labels and observation dates. Missing values remain unavailable. The new service does not imply that every listed platform exposes automated metrics or that repeated usernames prove common ownership. Public adapters refresh hourly; the page checks for published updates every five minutes while visible.
The browser also compares dates per platform, metric and demographic group, so a newer GitHub refresh cannot downgrade a newer Instagram snapshot already in the page. The initial service passed 15 unit tests and eight responsive/browser checks with no detected accessibility, overflow or console failures. The site release audit, Worker tests and all 45 feature/delivery/feed checks passed before publication. The first end-to-end workflow successfully published seven sanitized platform snapshots to Social Poller without transferring platform credentials.
Production verification passed all 26 canonical routes, 394 HTTP/cache variants, ten media-kit behavior checks, and the separate service’s search and cross-origin widget checks. Robots.txt is byte-identical to the committed source. The final site unit suites passed 46 tests plus the Worker assertions.
The integrated media-kit mobile Lighthouse sample scored 100 Performance, 100 Accessibility, 100 Best Practices, and 100 SEO, with 1.54-second LCP, 30 ms TBT and zero layout shift. The unchanged press page sampled 99 Performance and 100 in its other categories. These are laboratory observations, not guaranteed future scores or evidence of indexing. The performance review found no new render-blocking metrics request and no estimated saving from changing the existing stylesheets, so the page’s rendering strategy was preserved.
Latest collection: 2026-09-08T02:16:40.216Z. Instagram: 6/6 metrics refreshed; TikTok: 3/7 metrics refreshed; YouTube: 0/3 metrics refreshed; Twitch: 1/2 metrics refreshed; Bluesky: 3/3 metrics refreshed; X: 0/2 metrics refreshed; GitHub: 4/4 metrics refreshed; Instagram demographics: 3/3 breakdowns refreshed. Missing metrics retain their original observation dates.
Regression checks
Keyless Twitch public counters
Social Poller now has a DecAPI adapter for Twitch followers and concurrent viewers. It resolves the account’s numeric ID, verifies Sammy’s pinned identity, and parses only documented aggregate counters. An explicit offline response means zero viewers; missing, malformed or failed responses never become zero. The registered profile refreshes every five minutes, separately from site deployment. DecAPI also caches responses, so these are near-live snapshots, not an instantaneous stream of events.
The media kit identifies third-party cached counters and excludes them from current audience totals after ten minutes without a successful retrieval. Source records, history graphs and requested widgets use the same observations. The existing authorized connectors, static content, original design and indexing rules are unchanged. No new YouTube/TikTok total-count source is claimed. The adapter passed 31 Worker unit tests, type generation and a deployment dry run. Its deployed API and saved follower history returned source-labeled observations; 26 responsive widget/browser cases passed. Site checks passed the build, 26-route audit, Worker assertions, 25 feature and 16 delivery tests, twelve media-kit behavior cases and eight design/accessibility cases locally and in production. Production also passed 394 HTTP/cache checks and the 26-page, three-user-agent indexability sweep. Live robots.txt matched its committed source exactly. The real media-kit relay and isolated browser showed the source-labeled Twitch count without page errors. Mobile Lighthouse measured 99 Performance and 100 for Accessibility, Best Practices and SEO, with zero layout shift. Public Twitch lookup was also verified using a different channel: other users can use the same read-only profile API and widgets without their own platform key. Permanent history and background scheduling remain registered-owner-only.
Original creator design restored
The media kit and press reference now use the existing copper/gold palette, cream text, shared font stack, bold uppercase headings and square-edged panels. The page-specific purple colors and rounded dashboard styling were removed. Content follows the site’s bounded width and reflows for narrow screens. Requested Social Poller graphs, rotating cards and tickers use the same warm theme without loading an embed before the visitor requests it. Approved bios, metrics, observation dates, media files and downloads are unchanged. This styling release does not add new platform collectors or fill missing data. The build, 26-page audit, Worker tests, 24 feature tests, 16 delivery tests, eight design/accessibility checks and twelve media-kit behavior checks passed. The separately deployed widget theme passed 20 responsive browser checks. After deployment, all eight design/accessibility and twelve media-kit behavior checks passed against production. The real metrics relay returned 200. All 26 canonical pages retained matching indexing directives, all 394 cache-response checks passed, and the live robots file matched its committed source exactly. The mobile Lighthouse sample scored 100 in all four categories on both the media kit and press page, with zero blocking time and layout shift. Media-kit LCP was 1.54 seconds; this is a laboratory result, not a field-performance promise.
The implementation adds tests for unavailable versus zero values, stale dates, partial failures, account identity, aggregate-only output, static HTML, and indexing headers. The 26-page release audit and 394 local fresh/cached HTTP checks passed, along with the Worker suite, 39 unit/delivery/feed tests, and nine targeted browser behavior checks. The local mobile Lighthouse sample scored 100 for Performance, Accessibility, Best Practices, and SEO, with 1.24-second LCP, zero TBT, and zero layout shift. These are individual laboratory results, not field metrics or an indexing guarantee.
The expanded metrics release passed the 26-page build audit and 19 unit tests, including aggregate privacy filtering and independent TikTok permissions. Its authenticated refresh confirmed all configured Instagram counters and demographic groups, TikTok video-sample averages, Bluesky and GitHub counters, and Twitch viewer snapshots. Account totals that still lack authorization remain unavailable.
Production checks passed all 394 cache-response variants, matching HTML/HTTP indexing directives across 26 canonical pages, redirect checks, and discovery endpoints. The live robots file remains byte-identical to its committed source. Browser tests now distinguish populated demographic charts from their intentionally hidden empty-state messages and explicitly scroll and decode media-kit thumbnails before screenshots.
The final production browser sweep passed all 26 pages with no detected failures across four viewport widths, two accessibility scans per page, no-JavaScript and failed-script fallbacks, navigation, images, console, and CSP checks. Ten focused media-kit tests passed for refreshes, failure handling, demographic charts, zoom, and printing.
Live mobile Lighthouse samples scored 100 Performance, 100 Accessibility, 100 Best Practices, and 100 SEO on both the media kit and press page. The media-kit sample measured 1.54-second LCP, 18 ms TBT, and zero layout shift. This records a single laboratory sample, not a guarantee of search indexing or future scores. All 42 feature/delivery/feed unit tests, the Worker suite, and the three built feed checks passed.