Stratus Social and clearer media-kit charts

A separate profile explorer, source-labeled charts, and platform-by-platform audience breakdowns in the creator site’s original style.

The media kit now separates reported audience counts by platform using a shared linear scale. Current follower and subscriber snapshots receive a bar. Unknown and expired values remain explicitly unavailable; a real zero is still a measured zero. This is not a unique-person audience estimate.

Available numbers and connection gaps

Stratus now uses stratussocial.mellozone.site for its main site, documentation and media-kit chart embeds. An operator-assisted TikTok consent handoff is being connected to the existing scheduled collector. It checks the account identity, requested scopes and all four account totals before replacing the working grant. Account totals remain pending until that new consent succeeds; this is not public self-service enrollment.

The overview now shows YouTube’s raw subscriber count, separately from the comparison scale and combined audience. A failed refresh no longer hides a previously observed GitHub count: its original date and older-snapshot label remain visible. Current TikTok video-sample views and Kick live viewers appear as separate, clearly labeled measurements, not substitute follower totals.

GitHub’s scheduled public-data reads now use the Actions job’s built-in token. The collector reuses valid published GitHub snapshots instead of letting a rate-limited anonymous lookup suppress them. No private repositories are shown.

TikTok video metrics continue updating while account-statistics permissions are requested. The connected Kick channel API does not return follower totals, and Facebook Page analytics do not apply to the confirmed personal profile. Those gaps remain explicit rather than receiving invented counts.

Charts in the creator site’s own style

The existing copper, gold and dark palette, approved biography, creator artwork, press assets and collaboration links are preserved. Metric and date-range selectors make it possible to inspect a particular statistic without loading every external widget on initial page load. Motion remains opt-in.

Historical charts identify their axis range and show first/latest observations. Different sources, measurement scopes, precision levels and sample sizes remain separate. Missing periods do not receive invented data points.

A separate Stratus Social service

Stratus Social starts as a separate read-only profile explorer and widget service, backed by the existing Social Poller collector. The supplied logo is preserved on a white panel rather than redrawn. A developer guide and connection coverage page explain the available reads and embed options.

GitHub, Bluesky and Twitch support public lookups. Other enabled platform views use already-published registered-creator snapshots; they do not imply arbitrary account access. Self-service account connections and live event streams are not part of this first release. YouTube displays are not enabled on Stratus while its retention and redistribution requirements are reviewed.

The supplied X/FxEmbed starter passed its 23 mocked unit tests independently. It has not been activated: response limits, identity checks and enforced collection/publication permissions still need integration review. No paid X fallback, crawler-block bypass, or credentials from screenshots were enabled.

Available data still varies by platform. TikTok video-sample averages do not establish a follower total; missing accounts and demographic groups remain unavailable until a supported source supplies them.

Follow-up: collection gaps and short-range charts

The seven-day history view now preserves hourly readings instead of collapsing today’s collections into a single daily point. Metric selection prefers available observations, with unavailable choices clearly labeled. Longer ranges retain daily readings and missing history is never generated.

A bounded FxEmbed adapter was added for Sammy’s explicitly enrolled public X profile. It supplies source-labeled public counts without paid X API calls, login cookies, or an arbitrary-profile scraper. This replaces the earlier unactivated-starter status above; it does not supply private account insights.

The YouTube collector now resolves the official handle instead of depending on an unverified hard-coded channel ID. Collection diagnostics distinguish missing configuration and permission failures without exposing tokens or raw responses. Key configuration and live-source availability are verified separately from code.

The supplied YouTube key was configured in encrypted Actions secrets after explicit approval. Public YouTube API totals use transient publication rather than immutable Git snapshots, expire after thirty days, and stay separate from combined follower totals. The same Git exclusion keeps community X snapshots removable if the source becomes private or unavailable.

Production checks verified multiple same-day chart observations, the source-labeled X counters, and the real embedded graph in the creator theme. All 27 canonical pages passed indexability checks across three user-agent strings. The media-kit mobile Lighthouse sample scored 98 for performance and 100 for accessibility, best practices and SEO, with no layout shift or JavaScript exception detected.

The supplied Twitch app credentials were verified and configured for scheduled collection. A replacement YouTube key also returned the confirmed channel’s three public totals. TikTok still needs profile/statistics scope authorization. Remaining connection gaps are not zero-valued statistics or claims of complete platform access.

The refreshed production pipeline now publishes YouTube’s three public channel totals and Twitch’s authorized viewer snapshot. Browser regression coverage checks the rendered YouTube/X fields against the relay and TikTok’s available-history default.

Platform content follow-up

LinkedIn has been removed from the statistics dataset and media-kit selectors. This does not change the separate official identity links elsewhere on the site.

Platform content and account-type follow-up

LinkedIn has been removed from the current statistics dataset. Facebook is identified as a personal profile, with Page-only metrics removed. The Kick read-only connection has been tested; its current channel response reported zero live viewers, not an unknown follower total.

Platform content panels now support recent-post samples, per-post counters, ranked bar charts and observed gains when two compatible readings exist. Samples are capped at twenty records, rankings show up to ten, and empty or missing data is not substituted with estimates. YouTube displays raw counters without a derived engagement score. The scheduled collector uses expiring backend storage rather than committing post-performance samples to Git.

Reddit access is still awaiting an approved connection. The Bare Devvit starter is a separate Reddit-hosted app, not an external Data API token. TikTok video access works; account totals still require the additional approved profile/statistics scopes and renewed authorization. The first production content refresh passed: twenty TikTok posts, twenty Instagram posts, twenty YouTube videos, twenty X posts, nineteen Bluesky posts and nine GitHub repositories were collected. Twitch returned no retained videos; Kick was offline. These are dated samples, not lifetime totals. Live checks also corrected a Reddit handle-capitalization mismatch so its missing-connection state is reported correctly. Production verification passed across two real content collections, including TikTok post rankings and observed-change charts. The latest mobile Lighthouse check scored 98 for performance and 100 for accessibility, best practices and SEO. Canonical/indexability checks and 409 HTTP revalidation checks passed. Access-dependent gaps remain explicit: TikTok account-total scopes, personal-profile Facebook analytics and approved Reddit access.

Complete display of the collected data

Content panels now start expanded and read their cached sample automatically when they approach the viewport. Initial page rendering does not fetch every platform, and visitors never trigger upstream collection. Manual refresh remains available. The all-records view includes up to twenty collected entries, even when a date or ranking metric is missing. Field-by-field coverage distinguishes missing values from measured zeroes; unknown data is never filled with estimates.

Core reporting coverage and available period reports are visible in each platform panel. Additional reports no longer distort the core connection status, and each eligible platform has a direct metric-history control. Missing Facebook/Reddit connections, TikTok account-total permissions and unavailable Kick follower totals remain explicit. These display improvements do not grant additional data access.

A new end-to-end gate checks all ten platforms for unmapped ingested fields, API-to-card agreement, every collected content record and its raw metrics, demographic coverage, responsive layout and browser errors. The local-to-live coverage check passed all ten platforms and all 108 records in the current collected samples. The build, 27-page release audit, Worker tests, 43 feature tests, 16 delivery tests, content-chart accessibility/reflow checks and 16 media-kit behavior checks also passed.

The deployed page passed the same ten-platform coverage check, real history-chart checks and 409 HTTP/robots revalidation checks. Its mobile Lighthouse sample scored 99 Performance, 100 Accessibility, 100 Best Practices and 100 SEO, with 1.69-second LCP, zero total blocking time and zero layout shift. These are measured release results, not a guarantee of future scores or complete access to every platform’s private analytics.