Repository navigation
P0: verify appdevelopsk.com growth changes live and GA4 before acquisition rollout #51
Description
Why this remains open
2026-11-05 public read-only check of https://appdevelopsk.com/ confirms the previously merged count-neutral hero and featured order Debate Me → Tappitsu Battle → FutureMe are live. However, the just-merged related-app links from PR #50 have not yet been confirmed on public /support/debate-me/ (the checked copy lacked the new section at the time of the read). A successful GitHub merge does not prove full live propagation, browser functionality, or any install increase.
Already in main
- feat(acquisition): promote top apps and refresh 17-language homepage copy #39 homepage acquisition candidate priority/17 language
- feat(seo): add FutureMe/Tappitsu schema on latest main (clean extraction) #43 FutureMe + Tappitsu landing JSON-LD
- fix(growth): complete priority-app click attribution and accurate catalog metadata #45 accurate catalog count + per-app GA4 outbound click tagging
- feat(growth): 17-language related-app discovery and measured internal navigation #50 related-app discovery +
related_app_click+ sitemap lastmod
Operator verification — no GitHub Actions
- Open the three public pages in a fresh private browser: /support/debate-me/, /support/tappitsu-battle/, /support/future-me/. Verify a visible, usable two-link related-app section with no self-links, mobile layout and keyboard navigation.
- Confirm each new destination resolves successfully. Check English, Japanese, Arabic RTL and at least one other locale; then audit all 17 translation headings. No invented store availability.
- Check browser devtools console and GA4 DebugView for one
related_app_clickper click with only source_app, target_app, placement; one existingdownload_clickper real Play click with app_slug and placement. Confirm that outbound store clicks are not counted as installed users. - Confirm homepage title, catalog descriptions, JSON-LD and live sitemap entries reflect current source. Google search cache may be older than live HTML.
- Confirm how appdevelopsk.com is deployed and preserve deploy logs/commit SHA and rollback plan. If live pages lag main, do an authorized manual site publish; do not assume GitHub merge is a deploy.
- Compare GA4 7-day new users, landing-to-store click rate, store first-time acquisitions and matured D1/D7 by actual release date/country. Separate prior overall new-user trend from causal attribution.
- Verify authentic app-store screenshots, localized listings and real release version before new ASO tests. User handles App Store Connect/Play Console and mobile signing.
No GitHub Actions, no store publication, paid placement or fabricated campaign results. Close only after public/browser analytics evidence is attached.
Activity
appdevelopsk commentedon Nov 5, 2026
2026-11-05 update: source-side acquisition PR #55 was merged (catalog detail links and click attribution). Release verification script PR #56 was merged at fc32f07. To validate on a Mac without GitHub Actions:
This performs a read-only HTTP check of homepage, 92-card catalog, 3 priority support pages, robots and sitemap. Nonzero exit is a deployment/release gate failure, not evidence of traffic decline. Tests were added but not executed in this connector session. After live source passes, separately verify JavaScript and GA4 DebugView in browser, Play Console acquisition, store listings and mobile/RTL layout. Do not close until evidence of production publication and event delivery is attached.
appdevelopsk commentedon Nov 5, 2026
2026-11-05 additional completion: catalog app-detail links merged via #55 (549863f), smoke runner and six offline fixtures merged via #56 (fc32f07), and stale-live-catalog regression merged via #57 (0db1190). The Python suite now contains seven test functions, but has not been executed in this environment.
Public search retrieval today still returned old catalog copy ('92 free apps on Google Play') and no related-app section on one or more individual support pages; caching/deployment status requires verification. Run the source QA and live smoke locally:
Then check live browser/GA4 DebugView separately, keeping this issue open until evidence is attached.
appdevelopsk commentedon Nov 5, 2026
2026-11-05 21:19 JST follow-up: merged PR #58 (
90ae436ab747044b3ea538c0e4e4521448e8cbb8) strengthening the read-only live smoke: now checks catalog has 92 distinct Google Play Android package IDs, and verifies exact package-ID parity/count/order against JSON-LD ItemList. Added two more offline regression cases, for nine test definitions. Source inspection of current/appsfound 92 unique package IDs and expected UTM/referrer parameters. The regex/JSON-LD matching logic was exercised on synthetic good/duplicate fixtures; complete Python 9-test suite was not executed, nor was public HTTP smoke confirmed. Web retrieval today still produced a version of/appswith no new detailed-app section, and a version of the Tappitsu support page without its new hero CTA, so production cache/version verification remains a P0 blocking gate. No GitHub Actions or store submissions. Runpython3 -m unittest discover -s scripts -p 'test_growth_live.py' -vandpython3 scripts/check_growth_live.py, then verify the browser UI and GA4 DebugView, saving release SHA/country/time/window evidence.appdevelopsk commentedon Nov 5, 2026
2026-11-05 21:19 JST verification correction: The nine offline Python tests from merged PR #58 were executed successfully (9/9) in the ChatGPT local Python environment (not GitHub Actions). Both local file hashes match the exact blobs fetched from GitHub main:
scripts/check_growth_live.py=3515d64efa1c716fd29cd6bff8c5190d9b0b4ae4,scripts/test_growth_live.py=df41efc8b34760de1a0e0a4ad9f4da89567d2229. Tests cover baseline, missing page, missing related link, disallowed icon crawl, stale homepage/copy, duplicate Android package ID, schema mismatch, count mismatch. Actual public HTTP smoke could not be executed in this environment due to DNS/network resolution failure. Browser execution, deployment provenance, GA4 DebugView and Play Console measurements remain open. No GitHub Actions run.appdevelopsk commentedon Nov 5, 2026
2026-11-05 JST, acquisition closure evidence:
appdevelopsk commentedon Nov 5, 2026
2026-11-06 JST acquisition update: PR #59 merged (0990352), extending related-app referral web flow from three to five apps: Debate Me, Tappitsu Battle, FutureMe, Dotchi (repo slug
twotaku), Instant Judge.Code/source QA: 5/5 pages have 17 localized related-app headings, exactly two distinct non-self internal links, GA4 related_app_click/download_click hooks and both hero/detail CTA positions. /apps now has five detail spotlight links and still has exactly 92 catalog cards. Live smoke scripts expanded to five app pages and now contain 11 test methods; no Python test run claimed for this new commit. Sitemap retains 319 entries and updated lastmod for modified /apps, /support/twotaku/ and /support/instant-judge/.
Read-only public retrieval on 2026-11-06: /apps returned a result crawled today showing updated count-neutral description, but did not show the already merged app-detail spotlight; /support/instant-judge/ result was crawled last week and lacks related-app section; /support/twotaku/ fetch was inaccessible. This is incomplete/mixed-age crawler evidence, not a reliable deploy-status verdict. Execute the repository's read-only Python HTTP smoke against live and verify browser JS + GA4 DebugView before closing.
Outstanding: authenticated deployment/commit-to-live confirmation, responsive/RTL/browser checks, Play Console first-time installer and conversion evidence, and ASO/short-video publishing. No GitHub Actions used.
appdevelopsk commentedon Nov 5, 2026
2026-11-06: merged acquisition validation PR #60 at cdb54d0.
Fixed a live smoke false-negative: previously the presence of data-related-app / data-detail-app alone could pass even when href led to an invalid page. The checker now validates actual destination URL equals /support/{slug}/ and five distinct catalog-detail links. Added two corrupted-URL regression cases, bringing the suite from 11 to 13 tests. Source regex checked on actual 5 catalog links and Twotaku related links; the exact new Python unittest suite was NOT executed.
2026-11-06 read-only public retrieval: homepage showed the corrected priority feature order and count-neutral messaging; /apps result did not expose the new five-app detail spotlight. Instant Judge result was crawled the prior week; Twotaku and robots.txt could not be fetched by this retrieval. Mixed-age and failed fetch results are INSUFFICIENT to say a deploy failed; verify with authenticated publish evidence or run the included HTTP smoke on a networked machine.
Owner release gates remain open: exact main test run, HTTP smoke against live, mobile/RTL and browser devtools validation, GA4 DebugView (one related_app_click/download_click per real click), Play Console actual store visitors/first-time installers, ASO screenshot tests, creative publishing. Do NOT use GitHub Actions.
appdevelopsk commentedon Nov 5, 2026
2026-11-06: PR #61 merged into main as f5bc398. The live smoke previously required exactly 92 store cards. It now requires at least 92 as a non-regression baseline, enforces unique Play package IDs and an exact match with JSON-LD ItemList. This permits genuine 93+ app releases while rejecting partial catalog/schema updates.
Test proof: Python unittest executed locally on the exact PR file contents (Git blob hashes
36c32dd9491cd90a10e6962f81672a72acaa1b73and69b448fceb9e029e5cf7d39c2785ae47535c7867), 15 passed / 0 failures. GitHub Actions not used.Public read-only retrieval 2026-11-06: homepage shows corrected featured app order, catalog response is missing the 5-app detail spotlight, Instant Judge response lacks related app section; Twotaku and robots.txt could not be fetched by that source. These retrieval differences are not sufficient to determine the exact publish SHA, because source timestamps/crawling and deployment may differ.
Next owner gate: deployment commit SHA/URL verification and read-only live smoke on network-enabled Mac, then browser mobile/RTL, GA4 DebugView and Play Console first-time install conversion. No store publishing or SNS automation was attempted.
appdevelopsk commentedon Nov 6, 2026
2026-11-06 SEO release update: PR #62 merged to main at 478705a.
Added factual SoftwareApplication JSON-LD to /support/instant-judge/ based on existing Android Play package com.instantjudge.instant_judge; no invented downloads or ratings.
Extended source/HTTP smoke for all five priority support pages: canonical URL must exactly match the public page, disallow unintended noindex, and require parseable relevant JSON-LD referencing Android. Verified canonical and structured-data conditions in the five GitHub HTML source files. 18/18 offline Python tests executed successfully on exact Git blob hashes ca1be0c04410d756417f8e988e0c0dea99de58a8 and fac63f2a7733620f1541bb68d5ec8e3c6714c28b.
Public status remains OPEN: external search result for /apps on 2026-11-06 still showed no catalog spotlight in parsed response, and support/robots could not be fetched reliably. This does not prove nondeployment; record actual deployment commit hash, run authenticated network live smoke and browser tests before closing.
No GitHub Actions, no store-listing edits, no site deploy claimed and no installation lift claimed.
appdevelopsk commentedon Nov 6, 2026
2026-11-06 latest acquisition analytics fix: PR #63 merged at 3611409.
Found inaccurate GA4 semantics in homepage source: clicking an app card emitted
app_view, although no app-detail page view had yet been confirmed. Changed toapp_detail_clickwithapp_slugandplacement=homepage_card. GA4page_viewremains separate; this is a deliberate change in the event time series (historical app_view events were actually clicks).Live smoke now asserts homepage click event is properly named; two regression cases added. Exact GitHub blob SHA verified in local container, Python 20/20 tests PASS, 0 failures: checker 4c5b6db510bd54025a1628a3b418796e6dd314dc; tests a28fb4b215817adbf8aff968058c1c24b1b2fef5. GitHub Actions not used.
Live source verification remains BLOCKED/UNVERIFIED: public crawler fetched /apps today without the five-app spotlight, and Instant Judge today without related-app section. Crawled result is not the same as an authenticated deployment SHA or full live browser. Need deployment settings/logs, read-only HTTP smoke on networked Mac, JS event validation in GA4 DebugView, Play Console first-time-installer and store CVR.
appdevelopsk commentedon Nov 6, 2026
2026-11-06 growth release blocker and fix: PR #64 merged as 0361f7d.
All seven entry pages (/, /apps, and /support/ for debate-me, tappitsu-battle, future-me, twotaku, instant-judge) now contain identical
<meta name="sk-growth-release" content="growth-20261106-wave2">in GitHub main. Source read-only smoke now requires that exact marker in each page, so it fails if some pages still serve the old/cached release. This marker is a version label, not a Git SHA.Local unittest run 22/22 passed. Blob hashes identical to GitHub PR source: checker d55acbe1435ba0aa9549da1a218952a9913c33af and tests fa71b51c97ac651ed12bdd46c787b6d77939587c. Repaired catalog expansion fixture to preserve new metadata.
Independent live smoke against appdevelopsk.com attempted from container; failed due to DNS/network restrictions, not application response. Public read-only crawler responses of /apps and Instant Judge still did not show prior new sections; these are not authoritative verification of latest deployment.
Owner release gate: publish main code, inspect page source for growth-20261106-wave2 on all 7 URLs, execute
python3 scripts/check_growth_live.pyfrom connected Mac, verify 22 offline tests + actual browser clicks and GA4 DebugView, then compare Play first-time installers. Keep issue open until publication and behavior evidenced; no GitHub Actions.appdevelopsk commentedon Nov 6, 2026
2026-11-06 11:22 JST growth QA update: PR #65 was squash merged into main at e069944.
The read-only growth smoke now rejects mismatched Play Store package IDs across hero CTA, details CTA and structured data on each of the five priority app landing pages. Checked all five current GitHub HTML pages (5/5 package IDs agree). Three new regression methods bring the test count to 25.
Proof: Python unittest 25/25 PASSED on local code with exact matching GitHub branch blob SHA: scripts/check_growth_live.py = 24822797f8e5719be84114214bba063f707d48f7, scripts/test_growth_live.py = 267fba433d987508772859518be9510a57d99496. GitHub Actions were NOT used.
Public fetch via web on 2026-11-06 still exposed older catalog content without the five detailed-app links and an older Instant Judge page without related apps. This indicates a source-versus-public-content discrepancy in fetched results, but does not prove deployment failure (cache/crawl timing possible). robots.txt could not be fetched from this source. Open gate: authenticated deployment commit/version verification on all seven pages and live HTTP smoke, responsive browser checks, GA4 DebugView outbound-click actual delivery, Google Play Console install conversions and ASO tests.
Note: growth release meta value is a manual milestone marker, not a cryptographic content hash; source/content tests remain necessary to validate specific fixes.
appdevelopsk commentedon Nov 6, 2026
2026-11-06 JST: PR #66 merged as 1a205b8. Added a real zero-result recovery control to
/apps: clears search, resets category to All apps, restores all cards, focuses search, and tracks only non-PII catalog_search_reset. Preserved 92 store cards/5 spotlight links.Locally ran 27/27 Python unittest cases successfully on exact GitHub branch blob contents: checker a82647843fab435be068cf0ad0d8f1ffedae9abb, tests 99de2cb819f621b1af72efc6ca8de286adfb634e. No GitHub Actions.
Today's public read-only web fetch of https://appdevelopsk.com/apps still returned the older 'No matches. Try another word or choose All apps' empty state, and did not expose the already merged five-app spotlight. This does not certify deployment failure, because index/cache refresh and publication timestamps are unknown. twotaku support and robots could not be fetched by web source.
NEXT REQUIRED: deploy latest main through owner-controlled publishing and run Python HTTP smoke, compare all seven
<meta name=sk-growth-release>markers, then verify dynamic reset behavior in browser and GA4 DebugView. Measure actual Play Console store listing views and first-time installers; site-to-Play clicks are NOT installs. Keep this issue open until production evidence attached.appdevelopsk commentedon Nov 6, 2026
2026-11-06: PR #67 merged at 6e6079b.
Catalog search previously searched only visible app-card text, so searches for internal/former slugs such as twotaku (public app name Dotchi) could miss existing apps. Now the search corpus also includes image-derived stable slug and its hyphen-separated alias, without sending raw search terms to analytics. Preserved 92 Play cards, five app detail links, categories, and privacy-safe existing measurement.
Python regression: 29/29 PASSED locally on exact GitHub branch blob hashes check_growth_live.py 48f3e2539b781534064f54d10a548562c072445a and test_growth_live.py 3eedecc5c2c3e0919a6c6c92c77fbea05a52c081. No GitHub Actions.
Read-only public retrieval of https://appdevelopsk.com/apps on 2026-11-06 still showed older catalog HTML in its parsed response: no five-app detail spotlight or empty-state reset, despite both in latest main. Direct support/twotaku and robots were inaccessible to the retrieval. Treat live as UNVERIFIED, not proven failed deployment. Run smoke against user-controlled live origin from a network-enabled Mac, inspect deployment configuration/caches, GA4 DebugView and Play Console actual installs. Closing this issue requires production evidence.
appdevelopsk commentedon Nov 6, 2026
2026-11-06 19:33 JST: PR #68 merged as 57a9aa8.
Corrected misleading GA4
catalog_search_usedanalytics: previously triggered immediately on the first typed character, giving unfinished query/result counts. The app catalog still filters instantly on keystrokes, but now emits search_used only on Enter or blur, once per normalized query, without transmitting raw user search strings. Explicit empty-search reset clears the reporting guard. Note a measurement-definition break: pre/post-release event totals are not directly comparable.Proof: 32/32 local Python unittest cases passed against exact GitHub main blob contents: scripts/check_growth_live.py=71a9d2926f2378926b645466c95d5d0b63317bed; scripts/test_growth_live.py=89205a7df44b38d03c9a3883b10fcf5720fb5b01. Three negative regression cases added. Existing 92 catalog store links and 5 app-detail links preserved. GitHub Actions unused.
Latest public web retrieval of /apps (2026-11-06) still showed older empty-state and no five-app detail spotlight, while repo main has updated code. May reflect deployment lag/caching/tool retrieval, not an authenticated deploy verdict. Actual live source/JS/DebugView/Play Console conversions remain owner verification gates; do not mark published or claim acquisition lift.
appdevelopsk commentedon Nov 6, 2026
2026-11-06 JST production deployment blocker triage: merged PR #69 at 186c263.
Inspected the repository metadata (default branch main, admin access) and top-level deployment config paths: no CNAME, .nojekyll, _config.yml, netlify.toml, vercel.json or .github/workflows/deploy.yml in repository; README previously minimal. This does not identify hosting provider, deployment source, production branch or active deploy commit. Do not change DNS based on guesses.
Public web retrieval of /apps and /support/instant-judge/ on 2026-11-06 showed older HTML without recent detail links/related app section, whereas main contains them; source-versus-public mismatch remains unverified as to cause.
Added scripts/diagnose_growth_deploy.py to compare seven live HTML responses against a local checkout: HTTP status, SHA256 bytes, milestone release tag, per-page success/errors, JSON output. It flags non-200 even when bytes coincide, and warns that host-side minification/injection may make raw hashes differ. Added README deployment-runbook steps. The exact GitHub file blobs were tested offline: 7/7 PASS; source blob 4c88dfcdec8139a2a610ba2dba2ac20e49e4a0a2, test blob 74b1d9d0f14294f14f8f27c9285bfa90d41e26f2. No Actions used.
Owner verification command after checkout of the actual deployment commit: python3 scripts/diagnose_growth_deploy.py --root . --output growth-deploy-diagnosis.json; python3 scripts/check_growth_live.py > growth-live-smoke.json; python3 -m unittest discover -s scripts -p 'test_*.py' -v. Capture hosting dashboard production domain/project/source branch/commit plus JSON outputs, then validate GA4 DebugView and Play Console first-time installs. Do not claim site publication or audience growth until supported.
appdevelopsk commentedon Nov 6, 2026
Verification addendum for PR #69: combined existing 32 acquisition smoke regressions and 7 new deploy diagnosis regressions into one offline Python suite, executed 39/39 PASS, 0 failures. Exact GitHub blob hashes: check_growth_live.py 71a9d2926f2378926b645466c95d5d0b63317bed; test_growth_live.py 89205a7df44b38d03c9a3883b10fcf5720fb5b01; diagnose_growth_deploy.py 4c88dfcdec8139a2a610ba2dba2ac20e49e4a0a2; test_diagnose_growth_deploy.py 74b1d9d0f14294f14f8f27c9285bfa90d41e26f2.
To close this issue, still need source/deploy provider evidence: production project and branch, active commit, live HTTP smoke and byte-compare results, browser GA4 DebugView and actual first-time installs from Play Console. No GitHub Actions was run. Public crawler showed stale content but host configuration is not established.
appdevelopsk commentedon Nov 6, 2026
2026-11-06 additional deployment investigation: PR #70 merged as f4bf642.
Public web retrieval on 2026-11-06 saw /apps listing in newer wording (no old 98 free apps claim) but /apps.html with stale 98 free apps copy, crawled 3 weeks previously. These two crawler results are not a simultaneous uncached HTTP response and do not prove a live split or stale deployment. /support/twotaku/ and /robots.txt were inaccessible through the web retriever; not confirmed 404.
Added read-only scripts/diagnose_growth_routes.py to compare /apps and /apps.html, plus /support/twotaku/, /robots.txt, sitemap.xml. Captures HTTP status, final redirect URL, canonical, milestone marker, content SHA256, public cache-response metadata (Cache-Control, Age, CF-Cache-Status, X-Cache, ETag, Last-Modified, Server). Does not log cookies or content and does not modify DNS/caches. README explains manual usage.
9 new route diagnosis tests passed locally; combined with existing 39 growth/deploy tests, 48/48 passed on a local suite. GitHub branch test module SHA matched the local test module; diagnostic script was reviewed in GitHub for equivalent semantics but its local vs GitHub blob had nonfunctional source differences, so do not treat 48/48 as a GitHub exact-content test on every source file. No GitHub Actions used.
Owner next: verify production hosting provider/source branch/deployment SHA; on network-enabled Mac run
python3 scripts/diagnose_growth_routes.py --output growth-routes.json,python3 scripts/diagnose_growth_deploy.py --root . --output growth-deploy.json, andpython3 scripts/check_growth_live.py > growth-smoke.json. Inspect /apps vs /apps.html routing/canonical/cache and compare actual live revision, then DebugView and Play Console installs. No production publish/GA4 measurement or acquisition uplift confirmed.appdevelopsk commentedon Nov 6, 2026
2026-11-07 JST continuation of acquisition work. Merged PR #71 at 9283311. Four editorial app guides now link to /apps?goal=quick-games|habits|decisions|ai respectively, and the catalog safely preselects only one of these allowlisted categories. Original unfiltered /apps breadcrumb remains available; no raw search terms or PII added to URLs. GitHub source QA checked all four pairs, 92 Play listing cards, 5 app-detail links, and matching category buttons. Added four-method Python source test module; it was not executed because exact source couldn't be transferred to the isolated test environment.
Merged PR #72 at 1878c8c. Normalized two homepage internal /apps.html links to the catalog's declared canonical /apps. Source QA: two old links removed, no remaining /apps.html href on homepage, three total /apps hrefs. No external URL redirect or hosting configuration was changed.
Public /apps and /apps.html retrieval on 2026-11-07 still returns older empty-state copy, no five priority app spotlight. These are retrieved page results, not direct simultaneous host deploy evidence. Owner-controlled provider/project/source-branch/commit and routing/cache diagnosis are still needed; do not claim deployments or install uplift.
Available read-only commands on networked machine from latest checkout: python3 -m unittest discover -s scripts -p 'test_*.py' -v; python3 scripts/diagnose_growth_routes.py --output growth-routes.json; python3 scripts/diagnose_growth_deploy.py --root . --output growth-deploy.json; python3 scripts/check_growth_live.py > growth-live-smoke.json. Verify GA4 clicks and actual Google Play Console first-time installations after deployment. GitHub Actions not run.
appdevelopsk commentedon Nov 6, 2026
2026-11-07 JST update: PR #73 merged to main, squash SHA
3112a52aa8e3b68b28e6f413b0358ae719b926c5. Catalog now fires a privacy-safecatalog_filter_landingevent once on first render when landing via one of four allowlisted editorial-guide?goal=categories, with goal, result count and source=editorial_guide. Manualcatalog_filter_usedremains distinct. Source-level checks confirm all 92 Play cards and 5 detail links unchanged; new Python source-check test added (5 tests in module), but no claim of executed Python tests or verified GA4 delivery.IMPORTANT live status correction from public URL fetches crawled today 2026-11-07: BOTH https://appdevelopsk.com/apps and /apps.html now visibly include five app-detail links and the 'Clear search and filters' zero-result button (older reports saying these were missing are stale). Dotchi support page /support/twotaku/ now shows related links to Debate Me and Instant Judge. However /support/instant-judge/ in the same fetch still lacks a related-app section, despite this section being in main. Static HTML retrieval does not verify JavaScript behavior, exact deploy SHA, GA4 delivery, or installs. Continue investigating partial propagation/caching/hosting; do not claim all seven pages fully synced.
Available fleet baseline in appdevelopsk/app-infra apps/latest.json is generated 2026-11-05 07:14 JST, measurement 2026-10-29 to 2026-11-04: 117 apps, 7d 472 active / 354 new (excludes listed test devices). Revenue month-to-date in that snapshot 99 JPY. Debate Me 112 new 7d, D1 16.6%, D7 3.9% (n386); Tappitsu Battle 75 new, D1 3.2%, D7 0% (n31); FutureMe 41 new, D1 9.6%, D7 0% (n114). These are pre-release baseline figures, not outcomes of PR #73.
Next owner gates: verify which production deploy commit serves each URL, run live HTTP/route diagnostics with browser network access; check catalog goal=quick-games first render and exactly-one new GA4 event in DebugView, check cross-app related events, then Play Console first-time installers and D1/D7. GitHub Actions unused; no store or production deployment action asserted.
appdevelopsk commentedon Nov 6, 2026
2026-11-07 growth follow-up: PR #74 merged at
4c64c23debfa46fbb8a458eb9cfbdc4ceb783f6d.Fixed a test-only growth blocker: scripts/test_guide_catalog_links.py previously required exactly 92 catalog cards, conflicting with the existing live smoke min-baseline design. This would fail when adding a legitimate 93rd app. New test accepts >=92; added negative/positive regression cases for 91 and 93 cards.
Executed 7/7 Python unittest methods in an isolated local fixture with the test source byte-identical to GitHub branch blob
b013a70dee500d7fe78584b15ac283b18f984d63. Fixture HTML was representative, not a checked-out GitHub site; actual browser, exact 92-card source integration, and publish behavior are not thereby proved. No GitHub Actions.Current live web fetch on 2026-11-07 confirms public /apps and /apps.html show five app details and zero-match reset, and Instant Judge now shows related apps Dotchi/Debate Me. These observations supersede prior stale reports that those particular UI elements were absent. Do not yet claim the newest goal-landing analytics fired live; static crawlers do not execute JS or access GA4 DebugView.
Remaining production gates: real JS filter/deep link/zero-reset verification; GA4 DebugView catalog_filter_landing and other funnels; Play Console first-time installs and CVR; mature cohorts; full 117-app rollout. Snapshot apps/latest.json generated Nov 5 remains a pre-release baseline, not proof of an increase. GitHub Actions not used.
appdevelopsk commentedon Nov 6, 2026
2026-11-07 JST: PR #75 squash-merged into main, SHA f18a978. Fixes stale
?goal=URL after manually selecting another app-directory category or resetting to all apps. Uses history.replaceState to update only allowlistedgoalparameter while preserving existing UTM and hash; keeps the search input text out of the URL. Source QA: original 92 app-store cards and 5 detail links unaffected; guide source test module now has 9 methods, including two new URL-sync checks. Focused Node.js VM test of the exact helper logic executed locally (4/4 cases pass). Full Python source test and production browser test not executed. GitHub Actions not used. Remains: confirm live publication and category refresh/back behavior, GA4 DebugView guide landing/click, actual Play store visits and installs.appdevelopsk commentedon Nov 7, 2026
2026-11-07: PR #76 squash merged into main
7eeec68fb5dc10a25b69231b7a7836f3d48e534a.Corrected misleading
catalog_filter_landingacquisition attribution. Previously every filtered catalog link/apps?goal=...was reported assource=editorial_guide, including direct/shared/bookmarked traffic. Now only an allowlisted same-origin referrer path matching the selected category getseditorial_guide; suppressed referrer, mismatched guide, cross-origin and direct visits aredirect_or_shared. Raw referrer URL and raw search text are never included in GA4 event payloads.9/9 local Node.js reference-algorithm test cases passed. The Python guide source test was updated to 10 methods but not executed here; the Node test reproduces the classification logic and is not an authenticated full browser/GA4 test. No GitHub Actions.
Remaining evidence: deploy main, test actual browser referrer behavior and GA4 DebugView with a guide navigation and pasted filtered URL; confirm
catalog_filter_landingsource distribution, store visit and first-time install CVR in Play Console and app retention cohorts. No deployment or user growth claimed.appdevelopsk commentedon Nov 7, 2026
2026-11-07 growth update: PR #77 merged as
ff87f668ddc087de1c9602bebec93624931b083a.Fixed catalog search GA4 undercount: using the HTML search-field clear (×) and typing the same query again did not emit a new
catalog_search_usedevent, becauselastReportedSearchpersisted. Reset that guard on empty input, preserving Enter/blur deduplication and privacy. Added scripts/test_catalog_search_events.cjs that extracts/executes the actual search handler source from apps.html in Node VM. All 6/6 locally passed on a snippet matching GitHub source; test script Git blob SHA7b66ad289a328ffe6e11b66aa7903e5240ad6068exactly matched GitHub. Browser/network/GA4 real delivery not verified. README QA command updated.Connected GSC
sc-domain:appdevelopsk.comfree_recent_three_days (finalized 2026-11-02 to 2026-11-04, no comparison): TaleTime support URL 21 impressions / 0 clicks / position 5.05; Debate Me support URL 20 impressions / 0 clicks / position 8.3; querydebate me18 impressions / 0 clicks / position 8.5. Samples are too small to infer true CTR or post-release improvement.PR #78 merged as
c0790cac703f183ee08d6092dc295c7ff176d7fa. Updated TaleTime and Debate Me titles and meta/social descriptions to accurately convey documented capabilities. Debate Me schema.description parity, canonical, Play ID and both CTA links preserved; TaleTime has no confirmed first-party store URL, so none was invented.Connected HYPD Analytics has accounts/properties available but no GA4 properties selected for MCP access; user must select an authorized property at https://app.hypd.ai/sources?utm_source=mcp&utm_content=none_selected to retrieve live GA4 data. Do not claim real event counts/install lift without that plus Play Console data.
Still unresolved owner-only gates: verify actual deployed main commit and live JS, GA4 DebugView, Play Console first-time installers/CVR and mobile device tests. GitHub Actions not used.
appdevelopsk commentedon Nov 7, 2026
2026-11-07 JST — GSC-driven app acquisition work completed in repository main:
download_clickplacement=hero/details + app_slug.app_slugandsource_app=face-reaction-battle, keeping legacyappparameter.pawtrace15 impressions/0 clicks/avg pos 9.13. Face Battle query 14 impressions/1 click/avg pos 6.36. This is low-volume pre-change evidence only, NOT post-release improvement.scripts/test_gsc_support_content.pyhas 7 new standalone source regression methods after PR fix(analytics): standardize app_slug on Face Battle alternative Play clicks #80, but has not been run against a full local checkout; do not mark unit tests green. No GitHub Actions.python3 -m unittest discover -s scripts -p 'test_gsc_support_content.py' -v, browser test both pages, GA4 DebugViewdownload_clickand Play Console installers/CVR. GA4 HYPD access was connected but no property selected as of last check. No store listing mutation, SNS posting, install or revenue growth claims.appdevelopsk commentedon Nov 7, 2026
2026-11-07 JST — SEO crawlability / release verification update: PR #81 merged into main as
0bc6e0a1e91d3831012ae44f664d105c811e7f61.Updated sitemap
<lastmod>for exactly ten verified Nov-7-changed URLs (home, /apps, four editorial guides, TaleTime, Debate Me, PawTrace, Face Reaction Battle). Sitemap still has 319 distinct canonical URLs, none removed or duplicated. Reviewed fullmainsitemap and compared with previous source: only those ten date strings changed. Dates in future are not asserted.Added new check_growth_live smoke checks
fresh_lastmod_<path>for each of the ten pages, accepting later genuine updates. Added two regression cases: stale<lastmod>and missing updated URL. Addedscripts/test_sitemap_growth_freshness.pywith four standalone structure/date checks.Executed local Python tests using exact latest GitHub
mainscript bytes:check_growth_live.pyGit blob SHA2e29903c9db062b46c0556aed34413974db32cea,test_growth_live.pySHA365ed4ab3d5fe502837cf7af218ea9cb096cdc78. Combined existing release/route tests, ranpython3 -m unittest discover -s scripts -p 'test_*.py' -v: 50/50 PASS, 0 failures. This run did not include the new standalone sitemap XML test because the full sitemap XML was not locally materialized; the sitemap itself was checked through connected GitHub source, including complete URL uniqueness and exact ten-target diff. Distinguish script regression PASS from production HTTP / Google indexing.Production
https://appdevelopsk.com/sitemap.xmlcannot be verified through the current web retriever, so do not mark deployment or GSC recrawl complete. Host provider/source branch/deployed commit remain owner gates. No GitHub Actions used, no production deployment or store edits, and no observed traffic/install lift.appdevelopsk commentedon Nov 7, 2026
2026-11-07 JST, follow-up to #81: PR #82 merged at
b5705377ca7967c6bb4b131a6f55637551cc1644. Fixes the initial sitemap date test that required exactly Nov 7 and would fail on a later legitimate edit. Now accepts any validlastmod>= 2026-11-07 and forbids future dates; includes a positive Nov 8 / negative Nov 6 test.Verified exact GitHub script blobs locally: check_growth_live.py
2e29903c9db062b46c0556aed34413974db32cea, test_growth_live.py365ed4ab3d5fe502837cf7af218ea9cb096cdc78, test_sitemap_growth_freshness.pyd593d435166310c60e190a3449c4943d07575b50.Ran local
python3 -m unittest discover -s scripts -p 'test_*.py' -v: 55/55 PASS, 0 failures. IMPORTANT: 50 release/route/smoke cases use their own source-defined fixtures; 5 sitemap structural/date cases ran on a 319-URL synthetic fixture, because the actual GitHub sitemap bytes were not materialized in the isolated runner. Separately inspected the actual GitHub sitemap with connector: exactly 319 distinct canonical URLs, ten verified date-only edits, no change to other URLs.No GitHub Actions run. Live public sitemap could not be retrieved by the web provider; deploy commit, GA4 DebugView, Play Console install CVR and Search Console recrawl are still unverified. This test evidence does not prove deployment or SEO results.
appdevelopsk commentedon Nov 7, 2026
2026-11-07 (Claude Code): #71〜#82 を appdevelopsk.com へ配信 (publish_app_discovery.sh --apply, 58d77f3)。配信リストに support/taletime・support/dog-breeds-encyclopedia が漏れていたため追加 (#83 → main 0838e57)。本番10ページ+sitemap.xml が main と一致、check_growth_live 92/92、check_app_discovery_live PASS。陳腐化していた検査/テスト (#53/#63/#71/#77/#78 追従) も #83 で修正し node tests 52/52。GSC の再クロールと Play 実績は未確認 (11/12 に計測)。
https://claude.ai/code/session_01J5anYNmtTi1EKefKjB5aoV
appdevelopsk commentedon Nov 7, 2026
2026-11-07 JST — editorial-guide conversion & measurement wave. All three PRs verified merged into
main: #84d2e309b4e73012b4d853e072884d1d8abe91e40e; #8566be623c84abf5ffc2ce77695304f7c4d3afb8a5; #8675998be20d15d9fa42850ad08b56399e88ee9a3a.PR #84: Four editorial guides (quick games 5, habits 4, decisions 4, AI 4) now each link all 17 app cards to their first-party
/support/<slug>/feature/FAQ pages, with explicit app-specific accessible link names. All 17 target support pages were fetched and confirmed to exist. Existing direct Play CTAs and their tracking parameters untouched. Reusesapp_detail_clickwith app_slug, placement=guide_card and guide. Exact GitHub Python test blobbe3036515588621f4640e206fcedbd402399c7c3executed locally against representative 17-card fixtures: 3/3 PASS. Actual branch HTML separately audited: 17/17 correct app/store/detail matching.PR #85: Every guide store link now emits the legacy
guide_store_clickAND shareddownload_click(same click, distinct metrics), with app_slug, store=google_play, location=guide, placement=guide_card and guide. Neither event is proof of an app install; don't sum both as separate installs. Exact GitHub Node test blob2c68a19464b864aed75e8b99a96ffa2476a77935executed locally against representative guide fixtures: 8/8 PASS, including missing gtag safety. Actual GitHub source for all 4 guides was checked, each with one store handler.PR #86: Extends read-only
scripts/check_growth_live.pyPATHS to all four guides and verifies canonical, 17 expected links, store and detail click event definitions, app_slug, href/support mapping and accessible names. Updated existingscripts/test_growth_live.pyfixture and added 3 negative tests (total 37 methods in that module). Source and representative block verification: 29/29 positive checks and 4/4 injected faults detected. Do not report the full 37 Python tests executed; they weren't run in a fresh complete checkout in this turn.Remaining gate: user to publish current main via actual hosting project, then run
python3 -m unittest discover -s scripts -p 'test_*.py' -v,node --test scripts/test_guide_store_events.cjs, andpython3 scripts/check_growth_live.py. Browser/GA4 DebugView should confirm separateapp_detail_clickanddownload_clickevents and actual page delivery; Play Console first-time installers, CVR and post-release D1/D7 still unverified. HYPD GA4 property selection was missing on last check. No GitHub Actions, deployment or store-update actions in this turn.