Contest Readiness

Word Pairs contest readiness

1 November 2026 stopped-work record

The operator stopped product release work and deleted the recurring progress timer. The deadline report and supporting evidence record the exact-source local success, incomplete hostile reviews, subsequent staging migration/status fix and unfinished hosted release. This document's earlier checkpoints remain dated history. Later documentation refresh, wiki publication and Git synchronization do not resume product development, tests, reviews or deployment, and do not establish contest submission or requirement satisfaction.

Original audit date: 28 October 2026 UTC (27 October Central). Not yet cleared for submission. This is a release checklist linked to QBrain requirements, not an assertion of entrant eligibility or acceptance of legal terms. No entry was submitted by that audit. Preserve this audit and append later decisions rather than erasing earlier failures.

Canonical scope: Word Pairs / PROD-WORDPAIRS, current workspace, layer-1. The original readback receipt verifies QBrain TR-CONTEST-READINESS-041#CONTEST-01..05, TEST-CONTESTREADINESS-147 and their native mappings. A fresh effective-layer read through the Codex plugin on 30 October 2026 at 15:49 Central still found both records in progress and all ten combined criterion entries unsatisfied. QBrain remains authoritative; this document is their linked checklist.

Current release-scope amendment, 31 October 2026: the operator selected 2 through 10 players. Rules metadata targets v2.8 / 31 October 2026, creation default remains 4, observers remain 25, and maximum new-release audience is 35. The authoritative 19-record amendment preserves stable IDs and mappings; all 28 changed criteria remain unsatisfied. Existing larger running matches and stored results remain supported; oversized waiting/countdown rooms block new oversized admission/start without forced eviction and retain Leave. Follow the current completion plan. Earlier 15/25-player scopes and their failed/passing evidence are dated history. The exact-source local 10-player chain is now retained in the deadline evidence package; independent review, staged rollout and hosted observations remain unfinished; this amendment does not submit an entry.

Historical release-scope amendment, 30 October 2026: the operator approved 2 through 15 players, with the separate 25-observer limit preserved. The public game is guest-only, public sign-in and installation controls are disabled, and GitHub sign-in and PWA installation belong only to the private sharpninja administrator. Historical member profiles stay excluded from public results. Follow the current completion plan, requirements amendment readback and guest/admin release contract. Earlier 25-player, OpenAI sign-in and public-PWA records retain their historical scope; they do not establish the current release.

Official-rule register

Official rules, all seven pages; retained source PDF. This compact register is a navigation aid; the complete source governs.

Reference	Obligation / notice	Disposition
p1 Eligibility, Appendix A	Adult eligible-country resident, present there at entry; active Handshake account/entry email; excluded affiliations and households cannot enter.	Entrant attested all conditions; record in rules-receipt.json.
p1 How to Enter	Mission entry includes title, cover, description, URL; September22–October30,2026,23:59 Pacific.	Materials and manual entry pending.
p2 Rules of Entry	Multiple entries permitted; applicable law applies; eligibility/fraud information may be requested.	No automated entry planned.
p2–3 Selection	Execution, creativity, value, polish:25% each; top20 advance; top3 receive$1,000 each.	Evidence assessment below.
p3 Prizes	Valid notification email; possible eligibility proof, acceptance/release; taxes/costs belong to winner; claim conditions apply.	Entrant obligations, not verified.
p3–4 Information / Terms	Entry grants promotional submission/persona licenses and email contact; conduct code and OpenAI terms apply.	Entrant review pending; conduct text unavailable.
p4 Entry integrity	No tampering, harassment, mass/script/macro/automated entries; account holder is entrant.	Prepare locally; entrant submits manually.
p4–5 Legal / Privacy	Liability release, California individual disputes and damage limits, privacy policy; sponsor cancellation/disqualification rights.	Entrant review required.
p1–2 Purchase	Free entry; purchasing does not improve chances.	No entry payment requested.

Deadline history: the public brief says October31. The original audit selected the earlier PDF cutoff: 2026-10-31T06:59:00Z, October31 at01:59 Central, October30 at23:59 Pacific. Those source dates and the original decision remain historical evidence. On 31 October 2026 the operator directed: "Assume October 31." Use 31 October 2026 as the current working deadline, without inventing an exact hour or claiming sponsor clarification, an extension or acceptance of an entry. The generic submission guide calls preview images optional; the contest requires a cover, so include one.

Handshake account terms and submission flow

User-supplied Terms of Service, displayed update July22,2025; retained HTML. Applicable account/content sections require truthful identity; lawful, non-abusive, non-infringing content; necessary rights/permissions; recording consent; secure credentials; no unauthorized accounts, scraping, spam or misleading affiliation. Uploaded content is nonconfidential and licensed for service/promotional use. Machine-learning input excludes personal identifiers; generated output requires accuracy evaluation. Competitive misuse is prohibited. Enforcement, disclaimers, liability/indemnity, California law, arbitration/class-waiver/claim deadlines and entire-agreement provisions require entrant awareness. Employer/recruiter/paid-service clauses are conditional, not newly imposed game features. Special additional terms can control conflicts. This page does not identify itself as the Challenge Code of Conduct.

Project checks: keep generated guest identities separate from the private GitHub administrator; exclude old member and admin profiles from public results; exclude tokens/private logs/third-party recordings from entry evidence; audit asset licenses and content; never scrape Handshake profiles or automate account/entry creation. This audit performs no account changes, terms acceptance or contest publication. Existing moderation/ACL tests support their stated game controls but do not prove every future content item lawful. Entrant has attested eligibility. Licenses and final package review remain open.

The submission guide requires completing the Build steps before submission is enabled and using desktop. It distinguishes profile-only from AI Showcase submission. Verify the contest-specific path and required checkbox in the authenticated mission; a generic profile tile/badge is insufficient evidence of contest receipt. Retain the actual contest confirmation and timestamp. The Showcase overview describes truthful project/tool attribution and excluding confidential information; it does not replace this contest's entry instructions.

Incorporated OpenAI and privacy references

OpenAI Terms of Use displayed effective January1,2026: accurate account information, protected credentials, lawful inputs/outputs and input rights; evaluate accuracy, disclose AI involvement, respect intellectual property and service safeguards. Content ownership does not clear third-party rights. Regional or business terms may apply depending on the entrant account; billing, suspension, liability and dispute provisions remain account obligations. Account/region/contract applicability is unverified.

The sharing policy calls for review before sharing and clear AI attribution. Preserve accurate tool/author attribution in the description and release materials. The service terms describe feature-specific obligations and preview/beta limitations; do not promise permanent Work availability or infer indemnity.

For usage policies, use the October29,2025 revision effective on this audit date. The current landing page displays a November20,2026 effective revision, after the contest cutoff. Applicable safeguards address harmful/illegal activity, privacy, minors, deception, and sensitive decisions. No rule bypass or unsupported safety certification is claimed.

Handshake privacy policy displayed July3,2026: account/content/usage data, sharing and privacy choices belong to Handshake's service. It does not replace the game's own data practices. Use isolated guest evidence and inspect the final publication package for personal data. Public sources were read through the web tool; direct OpenAI HTML downloads failed and no retained-byte hash is claimed for them.

The operator confirms no other conduct links are available. The PDF's conduct and integrity provisions and Handshake account/content rules remain recorded. The separately named Challenge Code of Conduct and authenticated mission are still unavailable, with no missing-text compliance assertion and no repeated request to the operator.

Current product and evidence checklist

The public brief calls for a reusable multiplayer game at a public URL, room-code entry, synchronized screens and repeat play on phone/laptop without mandatory login or installation. It names ChatGPT Work. The current project decision is guest-only public play and a separate private admin PWA; the brief is not evidence that OpenAI sign-in is required. Codex usage alone does not prove the named Work workflow.

• Historical 30 October baseline. The 30 October serial validation receipt binds typecheck, lint, build and local tests to source b8400dc6f7aeb42a991861ac98e36f297eae8bb3. Its native reports contain 632 unit cases, 61 PostgreSQL cases and 79 browser cases, each with zero failures and skips. These are local source checks with disposable PostgreSQL and browser fixtures. They do not establish actual Realtime delivery, the player-count UI matrix, hosted behavior or current hostile approval. They remain history, not the current candidate's results.
• 31 October candidate and failed gate. The observed repeat, run 27a362dc-4720-4116-8640-a1f2016ab2c7, binds source 0e8d31a638e4db63e96302ca60fa7512c6accc96. Its source/build lane passed typecheck, lint, build, 894 unit/component/contract, 63 PostgreSQL and 114 browser cases. The PowerShell and actual Realtime lanes also exited 0. The normal UI matrix reports 13 passed, 1 failed, 0 skipped: rooms with 2 through 14 players completed eight rounds; 15 players timed out awaiting the category command's rendered acknowledgement before any solve. Strict UI and all four evidence-preparation operations did not run. The earlier failed 11-player chain and separate passing 11-player diagnostic remain separately retained. No Cline acceptance, hosted deployment or current release completion follows from these local results.
• Guest room-code play and public boundaries, CONTEST-02. Identity and room cases and the user-confirmed earlier local guest flow support their recorded candidates. At the final URL, repeat clean guest creation and code/invitation entry in two independent external browser sessions. Confirm guest navigation exposes no admin controls, public provider sign-in or public installation controls; confirm retired account routes cannot restore a public member session. Verify whitespace and the exact Copy Invitation Link control, successful clipboard content and truthful copy-failure feedback. Record the actual hosted candidate and observations.
• Synchronized complete games, CONTEST-02 and LIVE-07. Follow live recovery design, UI implementation and the current completion plan. Fresh actual browser integration must complete eight rounds at every count 2 through 10: 9 games, 54 player participations and 432 accepted round solves, with readiness, category choice, each answer and equal final rankings observed in the UI, at most two simultaneous answer submissions, and verified bounded cleanup. The scale matrix retains ordinary browser safety polling while using actual private Realtime and an owned TLS game database. It does not alone prove push-only delivery. Keep the separate strict two-player/observer case with periodic polling suppressed for push, reconnect and token-renewal evidence under LIVE-02/LIVE-06, plus the actual Realtime integration lane. Unit/SQL and HTTP-fixture tests cannot substitute for these lanes. Historical 11/13/15/20/25-player failures are retained with their distinct causes and unknowns; they are outside the current player cap, and the observer cap remains 25.
• Stable public address, persistence and replay, CONTEST-02. The deployment plan and read-only hosting preflight describe staging wordpairs and production wordpairs-j9jf; neither proves deployment of this candidate. Complete the reviewed staging-first rollout and pending privacy/capacity migrations. At the final production HTTPS URL, verify clean guest join, an eight-round finish, rematch/new room and persisted results across service restart. Retain mobile/laptop observations and verify that preserved legacy rooms/results remain accessible under the current capacity contract. Temporary tunnel history is development evidence only.
• Phone usability and private admin PWA, CONTEST-02. The guest/admin release contract defines responsive guest play without public install UI, a private admin manifest/service-worker scope and GitHub authorization restricted to the configured sharpninja identity. Check public phone/laptop navigation and authorized/unauthorized admin boundaries on the hosted candidate. Old member and admin profiles must remain absent from public rankings even if historical stored flags are public. Browser OAuth fixtures and Chromium installability checks do not establish real GitHub consent, physical OS installation or standalone-phone cookie continuity. Record those observations separately; installation is never required to play the guest game.
• Work provenance and mission checklist, CONTEST-03. The original audit could not verify the shared Work-chat content. A later retained follow-up records visual inspection of an accessible planning/design conversation whose final inspected response predates implementation. This qualifies the original failed read; it does not establish later coding, every response/attachment, or the authenticated mission's completion and submission instructions. Retain actual subsequent build evidence and inspect the mission when accessible. Neither tool usage alone nor a generic Showcase badge establishes those requirements.
• Release integrity, CONTEST-05. Preserve the GPL-2.0-only package, attributions, source-archive/SWBOM and HV archive. Generate and inspect the exact final release source/SWBOM, cover/artwork rights and safe evidence package. Exclude tokens, private profiles and provider credentials. These are project safeguards, not invented contest clauses. Use the current independently reviewed source and endpoint receipts; historical approvals cover only their recorded candidates.
• Submission materials and manual action, CONTEST-04. Use the title Word Pairs, a designated licensed cover, an accurate description and a durable playable URL. Follow the 31 October working-deadline instruction above while preserving the original rule-date conflict and checking the actual contest path. The eligible entrant reviews legal acknowledgements and manually submits; retain the confirmation and timestamp. This checklist does not authorize automated entry, accept agreements or establish that an entry will be accepted.
• Unknown mission/conduct content, CONTEST-01/03. The separately named Challenge Code of Conduct and authenticated mission/checklist remain uninspected. The operator supplied the official rules and account terms and confirmed no other conduct link is available. The exact missing dependency is access to that incorporated content and the actual contest submission path, not another request for the same unavailable link. Preserve the known clauses, record any later authoritative text, and do not claim full compliance while these remain unknown. Generic account or Showcase terms do not fill that gap.

At the current 31 October cap10 amendment, fresh actual Realtime/UI evidence, independent review and hosted verification remain release obligations. The current reviewer selection is Cline with model Beam and High effort until Monday, 2 November 2026; the exact cutoff hour is unspecified. A selection receipt is not a launched review or approval. Preserve all earlier Claude findings and other superseded reviewer selections with their original identities and outcomes.

Judging preparation

Execution evidence must show a complete stable group game. Creativity/value evidence should demonstrate the relationship-pair mechanic and intended social experience. Polish evidence should show clear onboarding, accessible feedback, responsive layouts, recovery, and safe moderation. These are recommendations against the published rubric; no winning score is claimed.

Evidence retention and final check

Keep all historical hostile-validation receipts, including rejected and superseded rounds, linked with hashes in hv-index.json. A historical approval covers its recorded candidate only. Preserve current failures until a later independent review explicitly verifies repairs. At release, freeze commit/tree, source manifest, dependency/SWBOM, run logs, original rules hash, cover and description hashes, final URL and manual submission confirmation/time. HV is supporting evidence; inspected public rules do not list it as a required upload.

Run this checklist again immediately before manual entry. Record changes to sponsor rules, resolve unknown mission/conduct text, reconfirm eligibility at entry, final origin and latest source evidence. Do not auto-check Handshake progress or automate entry; no rule acceptance or submission is authorized by preparing this package.

Eligibility update: operator answered "I meet all eligibility conditions" on28October2026UTC. This resolves the requested attestation; no independent identity proof or legal eligibility adjudication is claimed. Keep the entry account/email accurate and reconfirm residence/presence at actual entry.

Wiki publication is tracked separately by TR-DOCS-WIKIPUBLISH-042#WIKI-01..03 / TEST-WIKIPUBLISH-148. The original publication receipt binds its dated export to published wiki pages, assets, link transforms and verified remote commit; later release publication has a separate receipt. Re-export and verify any later amendments before claiming the wiki matches them. Publication does not submit the contest entry.

Category artwork update: the prompt/provenance manifest records the approved comic concepts and all twelve recomposed 9:16/16:9 assets, their hashes and generation prompts. Artwork validation and scoped hostile review retain the corrected full-page-background design and superseded review findings. This is artwork provenance and bounded UI evidence, not a contest submission or legal-rights certification.