September 8: live Discord follow-up

After the earlier browser-tool connection became unavailable, the operator resumed testing in authenticated Chrome through Patchright on September 8, 2026. The recorded interactions used the designated Vesperan application and test destination. This page summarizes completed observations through 06:52 UTC; identities, destinations, raw interface captures and interaction details remain private.

Both /play and /vesperan play returned the existing character. The live native menus exposed all ten groups and all 97 action labels, matching the current source. All nine legacy sections also rendered their actual idle bodies. These observations establish menu reachability; only the selected transactions below were carried through their recorded outcomes.

Recipe search reached Cook barley rations at catalogue position 69, beyond the first 25 entries, retained the selected recipe in review and returned through cancellation. At a 652-pixel viewport, the recipe picker showed two distinct pages of 20 options, Previous restored the first page and Cancel returned to the idle panel. The expert form selected the work method before exposing exactly the three Foraging goals. These checks cover the observed search, pages and profession; other recipes and professions retain separate acceptance work.

All twelve rules of a long policy were preserved across two review pages and a gateway restart. Commit was absent until every page had been opened, then remained available when revisiting the previous page. The operator cancelled that proposal. An unexpired native review continued after restart; old legacy controls instead gave their restart recovery instruction.

The legacy expiry check used an old menu after 935.75 elapsed seconds, beyond its 900-second lifetime, and recovered through a fresh command. The native ticket was refused after 3,795.38 elapsed seconds, beyond its 3,600-second lifetime; a fresh /play restored the same character. The native expiry and recovery were recorded at 06:47 UTC.

One routine foraging batch and one trail-ration craft ran on the live clock with 600-second and 120-second commitments. The craft was still running when the gateway restarted and remained visible afterward. Final carried staples were one Hedgeberry, one Bitterleaf and two Trail rations; refreshing and checking through the other entry point preserved those quantities. This is observed starter production and restart recovery, not a repeated-commit or full-ledger replay test.

The operator reviewed actual Discord screenshots. At 652 pixels wide, the resized desktop client supported the recorded recipe paging and cancellation path. The long policy reached its second page and all twelve rules were reconstructed from the page text, but cancellation by pointer or keyboard at that width remains unverified: repeated attempts had no recorded interaction or expected response. At 06:52 UTC, a programmatic click of the visible Discord Cancel control cleared the proposal, returned to idle and preserved the starter inventory; no policy was committed. Valid viewport captures show dense review JSON; two unsuitable locator captures were excluded from visual acceptance. At 390 pixels the sidebar hid the chat. These bounded desktop observations do not establish general visual, native mobile or accessibility acceptance.

Some helper and dropdown attempts lacked a recorded interaction or expected response. They remain recorder/client observations without a confirmed bot failure. Retained successful receipts define the completed scope; an incomplete observation or a changed selector label alone does not establish a finished bot response.

Player-copy cleanup remains: the Trail ration description includes laboratory and future-implementation wording despite its working live recipe. Broader native transactions, complex editors, multiplayer, endgame, sustained load and full Floor 1 acceptance remain open.

The earlier combined verification remains a separate dated record: 840 test methods, 3,894 subtests and 232 isolated Chromium checks, with 268 unchanged source inputs. This live follow-up does not add its observations to those historical counts or turn automated evidence into mobile, multiplayer or release approval.