Select second client tester cohort for availability and testing experience

Organizer confirms the ten-person round has already been invited. Do not expand or change its membership; any new Mac invitations are a separate two-person mini-round. Refresh and distinguish new response IDs from the last reviewed 52-response snapshot. Prepare a separate Mac mini-round of two after the existing ten were invited. Preserve the ten-person roster and organizer exclusions. Refresh Mac applicants and propose two prioritizing realistic availability and testing/reporting experience; model diversity is secondary. Do not count first-round Mac tester Trepso as a new invite. Offered Intel Macs do not automatically count as supported native client targets. Preserve prior contact files and generate a separate mini-round export. Include Caloss2 (response 30) by explicit user request because of his known Twitch presence; this is a user-supplied reason, not verified test expertise. Exclude first cohort response IDs 26,41,25,28,15,49. Keep AO tenure and character level distinct from testing competence. Inspect current schema and safely project relevant testing-experience prose or report-reference indicators without exposing names/emails. Existing hardware-focused proposals are historical, not approved membership. Internal Windows Legion Go coverage already exists. Preserve the first cohort and its contact exports. Keep names/emails out of assistant context; private contact export only via code if needed. No invitations or production mutations; intended timing is a day or two, not a scheduled action. Historical hardware-focused proposal (superseded): prepared ten candidates from a pool of 52 (30 Windows 11, 7 Windows 10, 11 Linux, 4 Mac). Proposed response IDs: 45,39,5,16,36,34,51,13,27,7. Four alternates: 23,14,48,11. Validated ten unique IDs, no overlap with first cohort or alternates, exact current handles and diagnostic consent. Five adapter tests pass. Reused the contact-export and spreadsheet-formatting scripts, adding local filename/count/selection parameters so the first cohort files stay unchanged. Contact exports contain emails but were generated and checked within code without exposing contacts in tool output or assistant context. Spreadsheet has short hardware summaries and a separate full-spec/selection-notes sheet; XML/ZIP structure, row count and contact preservation verified programmatically, not visually in Office. Owner-only ignored artifacts: .dev-data/tester-selection/{selection-wave2.json,profiles-wave2.json,next-10-testers-private.csv,next-10-testers-private.xlsx}. First cohort selection and contact exports preserved. Membership remains a proposal; no invitations or scheduled actions. Confirm mpekas's reported GPU and secondary APU access, StoneVigil's SteamOS/Deck details, Splatt's actual driver branch, Decker's legacy audio/Windows 10 driver setup, and 6teeth.'s exact macOS. All ten declare at least 5-10 hours/week, seven declare 10+; these are not verified reporting quality. Availability/experience reassessment complete: 66 total responses, 14 new IDs (53-66) since last snapshot; previously captured answer fields on existing responses unchanged. All 14 new entries categorized. Current ten-person proposal after organizer exclusions: 30,53,34,20,6,35,61,57,55,65 (five new, five previously reviewed). Responses 56 and 62 are excluded and must not be re-added automatically; responses 35 and 61 are their respective proposed replacements. Selection reasons in invitation exports remain neutral; no disciplinary or reputational allegations copied into them. Caloss2 included by user requirement. Two declare 10+ testing hours/week and eight 5-10. Priority reserves: 39 (reproduction-focused, 10+), 61 (ongoing beta/reporting, 5-10), 9 (claimed NDA testing, 10+, clarify no-report statement), 54 (claimed professional test engineer, only 2-5), 60 (UI designer, 10+), 51 (PTR comparison, 5-10), 44 (MMO betas, 5-10). Expanded adapter with explicit opt-in filtered testing statements, availability details, played content and report-reference text/URL-presence flag; all actual URLs removed. No linked reports or claimed credentials independently verified. Do not conflate AO tenure, level or historical chat activity with QA competence; no new Discord scan performed. Current ignored owner-only files: .dev-data/tester-selection/{profiles-availability-refresh.json,selection-wave2-availability.json,next-10-availability-private.csv,next-10-availability-private.xlsx}. Spreadsheet main sheet now prioritizes availability, self-reported experience, proposed test focus and timezones; separate hardware/notes and all 14 new-applicant assessments. Emails exported and programmatically checked without printing/opening them into context. Eleven tests pass, ten unique IDs/handles/consent verified, no first-cohort overlap, required Caloss2 present. Workbook XML/ZIP, 11/11/15 rows and frozen headers validated; no visual Office inspection. Existing cohort/contact files preserved. Oct 31 Discord cross-reference completed separately; current selection includes identity-qualified participation summaries and guild join-date context, not trust or QA verification. Use next-10-current-private.xlsx/.csv; older exports are historical and contain superseded invitees. Stenohaline replaces Hartdir in the draft: 5-10 testing hours/week, claimed MMO beta/ongoing reporting experience, exact matched account, Jan 2025 guild membership and 67 messages on 3 days in the Oct 1-31 window (recent burst, not sustained daily activity). Chadswill remains the corrected exported handle. Twelve tests pass, including a negative test proving organizer-excluded IDs are rejected before any API read; final 10-person export excludes both organizer-excluded accounts. All 14 new-entry assessments retained; excluded applicant explicitly marked organizer-excluded without disciplinary details. Organizer confirmed Nov 2 that the ten-person round was invited; its roster is preserved and local status updated to invited_confirmed_by_organizer. Separate Mac mini-round preparation complete, not an expansion of that ten: refreshed 67 applications (one new since 66), seven primary-OS Mac applicants, six eligible after excluding Trepso. Proposed two: sebinox (54), claims test engineer/programmer, M4 Max/36 GB, 2-5 hours/week; verganas (63), M4 Max/48 GB, 5-10 hours/week and native-versus-CrossOver stability/performance comparison. Both M4 Max intentionally prioritize expertise/capacity over model diversity. Alternatives: Danimal (48, M5 Air/software engineer, 2-5), 6teeth. (13, M1/16 GB/4K AirPlay, no testing experience, 5-10), madcow (22, M2 Max/no specific QA, 5-10), 3Go (59, unspecified Mac, 2-5). Do not treat sebinox's offered Intel Mac as supported native client coverage. Oct 31 Discord data reused and explicitly dated; no fresh scan, credential/report verification or reputation claims. Mini-round files: .dev-data/tester-selection/{profiles-mac-miniround.json,selection-mac-miniround.json,mac-miniround-private.csv,mac-miniround-private.xlsx}. Private email export performed solely in code, not printed/read into assistant context. Twelve tests pass; exact handles/consent, two unique IDs, exclusion from invited ten/first six/organizer exclusions, disjoint reserves, XML/ZIP integrity and 3 rows per each of two sheets verified. No visual Office review. Separate mini-round remains proposed pending organizer approval/invitations; tools performed no invitations or production writes.