Localhost opens with no browser errors
Browser regression opens the dashboard at desktop and mobile widths and fails on console/runtime errors.
npm run m23:browserCovered by the browser script and latest local verification run.
This is the public checklist for the M23 site goal. It separates live evidence, automated test coverage, and documented evidence from account-owner blockers that still need external action.
Generated 6/4/2026, 10:42:49 AM
docs/m23-goals/10-acceptance-checklist.md
https://www.m23isgalois.com
npm run m23:verify
/api/m23/audit
Requirement evidence
Browser regression opens the dashboard at desktop and mobile widths and fails on console/runtime errors.
npm run m23:browserCovered by the browser script and latest local verification run.
The current dashboard snapshot has route, progress, verified units, pending units, and remaining units.
8/42 verified units; 34 remaining.
The starter work unit exists in the public queue.
m23-replay-0006 is available as the starter task.
The work-unit page has a ChatGPT Pro prompt path and the safety model treats transcripts as review-only.
The UI labels this path Use ChatGPT Pro and scripts verify the link.
The notebook guide and notebook API explain the task first, then produce the upload packet JSON.
The notebook path exists for m23-replay-0006 and is checked in browser tests.
Smoke tests post a complete result packet through /api/m23/results and expect a pending submission.
npm run m23:smokeCovered by API smoke tests.
Smoke tests reject secret-like content, and browser tests check the malformed JSON message.
npm run m23:verifyCovered by smoke and browser checks.
Smoke tests compare progress before and after a pending packet and require no completed-unit change.
npm run m23:smokeThe pending-does-not-count automation gate is green.
Smoke tests keep schema_valid from advancing progress, then require reproduced_same_script to advance it.
npm run m23:smokeCovered by verification-level smoke tests.
Leaderboard data is derived from submissions, reviews, and verifications; smoke tests require verified credit to appear.
npm run m23:smoke2 contributor profiles are currently visible.
The automation snapshot includes local verification, smoke, browser, production, strict, and GitHub stages.
6 automation stages are currently exposed.
The automation API exposes stage, gate, endpoint, and owner-action arrays.
6 gates; 2 owner actions.
Deployment, DNS, storage, automation, and strict-mode commands are documented.
See docs/m23-deployment.md and docs/m23-automation.md.
Readiness and automation snapshots list apex-domain owner actions when apex HTTPS is not verified.
Apex HTTPS/redirect has not been marked complete. Keep using the www URL until Porkbun/Vercel verification finishes.
Add or refresh the Vercel TXT verification records in Porkbun, then refresh Vercel Domains until apex HTTPS redirects to www.
Readiness and dashboard storage panels report whether production records are durable.
This deployment can demonstrate the workflow, but submissions are not durable until DATABASE_URL or BLOB_READ_WRITE_TOKEN is configured.
Create or connect Vercel Blob or Postgres, set BLOB_READ_WRITE_TOKEN or DATABASE_URL in production, and redeploy.
Route caveats, leaderboard honesty text, and verification pages state that unreproduced computations are not proof.
A completed orbit chunk is evidence for that chunk only. A large orbit or threshold exceedance is not a proof by itself. No single computation proves M23 is or is not a Galois group over Q. Later genus, Galois action, rational point, and descent gates still matter.
The public worker keeps AI endpoint credentials local and requires explicit acknowledgement before AI calls.
npm run m23:smokeSmoke tests confirm AI mode stops without acknowledgement and can use local endpoints.
Browser checks run dashboard, work, notebook, automation, leaderboard, contributor, submission, and upload paths at 390px.
npm run m23:browserCovered by browser regression checks.
The release command runs ESLint before build, smoke, and browser checks.
npm run lintLast local run passed; rerun before final release claims.
The release command builds the production Next.js app.
npm run buildLast local and Vercel production builds passed; rerun before final release claims.
Smoke tests exercise APIs for uploads, reviews, artifacts, verifications, progress, leaderboard, and AI/A2A safety.
npm run m23:smokeCovered by m23:verify and CI.
The browser regression script opens /m23/automation at desktop and mobile widths.
npm run m23:browserAutomation page checks are part of m23:browser.