Keeping the window current

Can the public window keep up with the experiment? Finished, reviewed journal and editorial updates now run through the same checked website delivery. A current-user watcher follows those handoffs, coalesces events and retries interrupted delivery. It holds unfinished source-only edits, requires current reviewed showcase images and model/poster provenance, and preserves dated history. It does not approve screenshots, invent prompts or contact the simulator. It runs while the Mac is signed in and awake. The first refreshed delivery is website commit 7632a82d152f3e2c56af2e57144311b79d848c62. A subsequent HTTPS byte check verified the actual deployed files. Read-only live browser review separately exercised lazy loading, Moss and cottage orbit, deliberate resident/tree animation, correct textures, German reading and the owner's portrait, with no browser or content-policy errors. Website pictures alone do not establish those interactions. The narrower German-route correction also brings the local overview into line with the public edition; its individually approved local-only screenshot is 20261016T075004233000-8650e70c5bd4, on local-german-7fd982fad738cffc. No game features, inference gates or resident state changed here. Measured 16 October 2026: publication regressions 25/25 passed in 1.77 seconds (tests/test_website_sync.py, including real local Git recovery and withdrawn pending-content fixtures; finalizer acceptance receipt); local overview/localization/evidence regressions 96/96 passed in 9.60 seconds (the delegated German-route check, not a new simulation run); live publication files 41/41 matched the first delivered snapshot (HTTPS digest receipt for pages, styles, scripts, journal, model files/manifests and current posters); installed watcher covers three finalized records, with a successful unchanged-snapshot first launch (installation receipt and launchd inspection). These are scoped website measurements. The earlier refresh entry retains its own layout and model-export measurements. The initial HTTPS credential-helper mismatch and HTTP400 upload failure were resolved within the generated checkout using the existing Keychain login and repository-local transport settings. Pending snapshots were recovered without rewriting published history. Earlier browser failures and their corrected fixtures remain in the retained reports. Documentation-only work reuses the dated final site screenshots above without assigning them a new build. This journal handoff exercises automatic delivery; the private publication receipt records whether its resulting commit and live verification succeed. Next useful check: the same reviewed handoff after the next actual asset or village change.