Nursing Edge

Service status

Current status

The Nursing Edge service is in progress

Daily updates

One entry per working day, newest first.

2026-11-18 · 0.0.31
  • Accessibility week (Wednesday): screen-reader labels and roles on learner Practice/Review question flow (MCQ/SATA fieldsets, confidence radios, support skip, end-session dialog) plus polite post-commit announcements that state outcome and confidence without reading full rationales
  • Admin document titles for Overview, imports, learners, audit, learning evidence, database, and access; filter panels and table FilterMenu controls have discernible names
  • Pre-commit player safety reaffirmed in player-question and player-load tests; notes and limits for Developer Complete in nursing-edge/scripts/accessibility-screen-reader.md
2026-11-17 · 0.0.30
  • Accessibility week (Tuesday): keyboard tab order and visible focus rings on learner Home, Practice, Review, and Progress shells, plus admin sign-in nav, Content Imports, Learners, Overview, Audit, and Database backup flows
  • Skip-to-main-content links on learner and admin app shells; global :focus-visible styling and scroll margin so focus is not hidden under sticky headers
  • Dialogs and confirmations (end session, question detail, database backup/restore, job overlay) trap focus while open and return focus to the control that opened them when closed; known limits are listed in nursing-edge/scripts/accessibility-keyboard-focus.md for Developer Complete notes
2026-11-16 · 0.0.29
  • Super Admin Database page: backup and restore the learner MongoDB database to private Vercel Blob archives (manifest collections, NDJSON, validate before restore, type backup name to confirm)
  • Backup and restore runbooks and the week-of-9-Nov security checkpoint now treat Admin Blob as the operator path; local mongodump and mongorestore scripts are optional engineering tools only
  • Shopify webhooks and production Vercel env (MongoDB, session secrets, Blob) confirmed by the client; security week gate closed with Developer Complete still ahead on the overall plan
2026-11-13 · 0.0.28
  • Friday security checkpoint (v5.1 Gates 26 and 27): dev and prod database naming, backup manifest, and a successful dev restore drill are documented; ops must verify MONGODB_DB_NAME before any destructive import or restore
  • Admin API CORS reflects only origins listed in ADMIN_ORIGIN; learner session cookies use Secure in production; Shopify webhooks stay rejected until SHOPIFY_WEBHOOK_SECRET is set and HMAC verifies
  • Unit tests cover pre-commit payloads, admin and learner route auth, and the checkpoint controls. Two invariants.test.ts delay cases remain expected until 23 November. Live Shopify revoke and restore proof is deferred until store credentials are on Vercel
2026-11-10 · 0.0.27
  • Unit tests lock pre-commit player payloads: open and held-feedback question JSON stays stem and safe options only; answer keys, rationales, and post-commit teaching stay off the wire before commit
  • Practice and Review pages load question content only through the trimmed server projection, not a full question version document
  • Every sensitive admin API route requires an active admin session; learner Practice, Review, Baseline, commit, draft, interview, and session control routes bind work to the signed-in learner, not a client-supplied session or learner id
2026-11-09 · 0.0.26
  • The learner API resolves MongoDB database name from MONGODB_DB_NAME, or from the database segment in MONGODB_URI when set, instead of always using a single hard-coded name
  • nursing-edge and nursing-edge-admin .env.example document local dev vs production names; nursing-edge/docs/mongodb-environment.md describes Vercel checks and when not to run destructive imports from a laptop
  • Admin UI still uses only the learner API for data; it does not open its own MongoDB connection
2026-11-06 · 0.0.25
  • Admin Audit lists publish, pause, delete, access grants, imports, link decisions, teaching decisions, and learner exports, with filters for actor, date, action, and outcome
  • Operator learner export is keyed on learner id. The file has attempts, Review cycles, and Interview deliveries, and leaves out name, email, and answer keys. Each download is recorded in Audit
  • Week-of-3-Nov gate checked in unit tests. One teaching decision records the admin, revision, and time. Imported teaching stays unapproved until Approve. Audit and export do not edit attempts, Review cycles, Interview records, or answer keys
2026-11-05 · 0.0.24
  • Admin Overview shows registered learners, active access, recent activity, Review due now, cycles near expiry, and backlog-style cycle counts. Counts only, no charts or scores
  • Admin Learning Evidence filters committed attempts by date, topic, format, confidence, six-state, session type, and Review outcome. Accuracy is correct divided by answered, with clinical and physiology kept separate, using the same countResults formula as learner Results and Progress
  • Both views are read-only. They do not edit attempts, Review cycles, or Interview records
2026-11-04 · 0.0.23
  • Admin Learners: search by email for a read-only snapshot of access, Calibration status, Practice and Review activity counts, and a timeline of committed attempts with confidence, six-state, Review effects, Interview deliveries, and physiology support skips as skips
  • Review Trace from a timeline entry shows the source attempt, rule set, stage, due and expiry window, pause extension, transitions, session seed and order, deferred owners, and when an approved link placed a sibling or physiology question
2026-11-03 · 0.0.22
  • Admin Content Imports has a Teaching tab. Approve or Ban records the admin, that exact teaching revision, and the time. Imported pre-commit teaching stays unapproved until you act; Interview opens only on an approved revision for the served version
  • Question detail on Content Imports shows permanent item, version history, publication, runtime role, and read-only learner usage: answered count, accuracy as correct divided by answered, confidence and six-state counts, and option selections by exact chosen set (including SATA)
  • Usage math matches learner Results: same countResults formula, not a separate admin calculation
2026-10-30 · 0.0.21
  • Home and Progress show What Happens Next from saved Review cycles: what is due now, the next due time, and cycles near the end of the review window. An ended window stays in history
  • Week-of-26-Oct gate checked in unit tests. Review still serves the original question while no alternate link is approved
  • Repair and Confirmation stay on the short test delays until 23 November
2026-10-29 · 0.0.20
  • Progress replaces the placeholder: current accuracy uses the latest valid answer on each clinical question, not every row in the database
  • First-to-current change uses the eight change classes from Tuesday's work
  • Evidence by topic lists all eight topics; a topic appears only after 10 clinical answers in it, with the same accuracy formula as Results
  • One topic accuracy chart. Physiology answers stay in a separate count
2026-10-28 · 0.0.19
  • Practice and Review ended-session pages show Results from that session's own committed attempts only
  • Assigned and answered are separate. Accuracy is correct divided by answered, not by assigned
  • Confidence counts, six-state counts, and accuracy by confidence reconcile to the number answered. Zero answered shows a no-answer state, not 0%
  • Practice and Review Results never merge. Review keeps clinical and physiology counts separate
2026-10-27 · 0.0.18
  • Eight first-to-current change classes with correctness first, then confidence direction on unchanged correctness
  • One comparison-compatibility rule for the same clinical identity: same question version or approved nonmaterial successor in lineage; siblings and physiology are never compared
  • Invalid, duplicate, and excluded attempts stay out of change and progress evidence
2026-10-26 · 0.0.17
  • Large workbooks import without timing out. Each sheet is checked and saved in one batch, and audit-only sheets are skipped
  • A timeout or failed upload shows a clear message instead of "Failed to fetch"
  • Each import result has a "Why N failed" list that groups failed rows by sheet and reason, with example rows
  • Workbook cells saved as inline text are read correctly instead of appearing blank
  • If Part 2 of a split workbook is imported before Part 1, its references to Part 1 fail as mapping errors while its questions still stage. Importing Part 2 again after Part 1 stores those maps. Nothing is published or approved
2026-10-26 · 0.0.16
  • Content Imports no longer depends on exact workbook file names. Split or renamed workbooks get the right role
  • A valid AP_ITEMS sheet marks physiology questions. Otherwise "Sibling" in the file name marks sibling questions, and any other workbook is clinical core
  • The import result shows the role used and any warning. Sibling and physiology questions stay out of Practice, and nothing is published or approved
2026-10-16 · 0.0.15
  • Practice leaves out sibling and physiology versions. A published question with no role stays eligible, and the Calibration 35 stay reserved
  • Content Imports has a Links tab. Approve or Ban records the admin, that exact revision, and the time
  • Imported workbook approvals stay null. Review still shows the original question
2026-10-15 · 0.0.14
  • Physiology workbook: 450 questions stage unpublished with role AP reinforcement
  • 450 mechanisms, 450 support links, and 5 variant groups are stored; support approval stays null
  • A physiology source-system name resolves through its topic binding. The five clinical workbooks keep their previous import path
2026-10-14 · 0.0.13
  • Clinical workbooks stage question content and store maps, directed links, topics, teaching, and evidence
  • 2,004 successor clinical versions stay staged as clinical core; 500 siblings stay staged and reserved from Practice
  • Published earlier versions stay published. Approvals stay null. A content failure does not remove a sound map
2026-10-13 · 0.0.12
  • Relationship collections and indexes for registries, maps, directed links, teaching, evidence, exposure, and roles
  • No question rows are rewritten, and nothing is published or approved
2026-10-13 · 0.0.11
  • Admin Content Imports: delete selected staged versions that have no attempts and are not on the Calibration set
  • A published or attempted version stays in the bank; one rejected id deletes nothing
2026-09-28 · 0.0.10
  • Admin can save a Product Owner Calibration set of exactly 35 imported question versions
  • Ordinary Practice eligibility excludes that Calibration set; the app does not guess the 35 from IDs
  • Upload as Calibration or Set selected as Calibration; a count other than 35 is rejected
2026-09-28 · 0.0.9
  • Admin Content Imports: pause selected published versions so they leave the eligible set
  • A paused version stays published and is not eligible until it is active again
  • Pause is fail-closed and audited; one rejected id pauses nothing
2026-09-28 · 0.0.8
  • Admin Content Imports shows an eligibility summary: imported, published, MVP-eligible, unavailable or invalid, and unsupported interaction type
  • The launch gate of at least 2,000 eligible questions is a report only and is not hard-coded into selection
  • Only published, active, valid, complete MCQ or SATA versions are MVP-eligible; a later session must use that same server list
2026-09-25 · 0.0.7
  • Admin Manual access: grant Nursing Edge access to an email that has not paid; no invitation is sent
  • A new learner receives a verification code only when that email is already paid or manually granted
  • After verification: Terms and Privacy, one-time orientation, then Home. Practice, Review, and Progress do not open questions
  • Admin Content Imports: publish selected staged versions; learners still do not see questions
2026-09-24 · 0.0.6
  • Learner register and sign-in with first name, email, password, and an 8-digit email code
  • A verified email that matches an unclaimed purchase links that subscription to the account
  • The account page shows whether access is active
2026-09-23 · 0.0.5
  • A signed Shopify payment creates one unclaimed entitlement
  • The buyer email is stored only as a hint. No learner account is created, and no card data is stored
2026-09-22 · 0.0.4
  • Checkout stays on Shopify. The app holds empty slots for the store and webhook secret and does not invent product or plan IDs
  • No question publish
2026-09-21 · 0.0.3
  • Admin Content Imports: upload a C2 workbook; each row is validated and staged, not published
  • Import page lists already-imported question versions for any signed-in admin after refresh or a later sign-in
  • Prior batches are collapsed by default and expand to show batch pass/fail counts
  • Learner content, import batches, and import lines stored in MongoDB; production reuses one database connection
2026-09-18 · 0.0.2
  • Admin register and sign-in with email, password, and 8-digit email code
  • Super Admin is created from the backend Super Admin email; other admins stay pending until approval
  • Admin accounts and codes stored in MongoDB (admin_users, admin_users_codes)
2026-09-17 · 0.0.1
  • Status site homepage with current service status and changelog
  • Learner and admin apps are separate sites