FEATURE LIST

Authentication

[x] Google Login
[x] Logout
[x] User profile (view only initially)
[x] Avatar from Google

School Hubs

[x] View school hub
[x] Search for school
[x] Request hub activation
[x] View departments
[x] Filter questions by department
[x] Manage departments (create/edit/deactivate/reactivate), School Representatives and admins

Questions

[x] Create question (title, body, school, optional department, tags)
[x] Edit own question
[x] Delete own question
[x] View question
[x] View question list (paginated)
[x] Filter questions by status (Open/Answered/Solved)
[x] Filter questions by school (via the hub filter, since a hub maps to exactly one school)
[x] Filter questions by department
[x] Filter questions by tags
[x] Follow a question for update notifications, independent of authorship (frontend not yet wired, backend endpoints exist and are documented)

Answers

[x] Create answer
[x] Edit own answer
[x] Delete own answer
[x] Mark best answer (question owner only)
[x] Vote on answers (upvote/downvote)
[x] View answers

Comments

[x] Create comment on answer
[x] Edit own comment
[x] Delete own comment
[x] View comments

Search

[x] Search questions (title + body)
[x] Search by school
[x] Search by department
[x] Search by tags
[x] Search results ranking (solved > highly voted > recent > relevance)

Notifications

[x] Email notifications for major actions
[x] New answer notification
[x] School hub activation
[x] In-app notifications for major actions
[x] New answer notification
[x] New comment notification
[x] Best answer selected notification
[x] Vote received notification
[x] Moderator assigned notification (assignment endpoint exists since Phase 14, no notification trigger wired to it yet)
[x] Hub activated notification
[x] New question notification for hub moderators (NEW_QUESTION, in-app only, closes a gap from project-plan.md's Moderator Responsibilities section, which described this but it was never implemented)
[x] New activation request notification for admins (NEW_ACTIVATION_REQUEST, in-app only) and new report notification for admins (NEW_REPORT, in-app only)
[x] Account suspended notification (USER_SUSPENDED, email and in-app, closes a real UX gap: a suspended user previously had no way to know why login stopped working)
[x] Hub activation rejected notification (HUB_ACTIVATION_REJECTED, email and in-app), closes the asymmetry where approval emailed the requester but rejection sent nothing at all
[x] Report reviewed notification for the original reporter (REPORT_REVIEWED, in-app only), closes the previously-tracked Platform Improvements backlog item

Moderation

[x] Report content (spam/abuse/misinformation/duplicate)
[x] View unanswered questions (moderator only, scoped to assigned hubs)
[x] Reports dashboard (admin only)
[x] Resolve reports (admin only)
[x] Question lock/close capability, distinct from SOLVED, moderator/admin only, for the rare case a question genuinely needs to stop accepting input (spam magnet, fully resolved administrative question). SOLVED itself no longer implies closed, see project-plan.md's Question Lifecycle note. Backend complete and documented; frontend UI for this is planned for the upcoming polish pass.

Administration

[x] Manage schools (create/edit/soft-delete via is_active)
[x] Approve hub activation requests
[x] Assign moderators
[x] Assign school representatives
[x] View reports
[x] Manage users (view/suspend; promoting or demoting admin status is not part of this endpoint)
[x] Manage tags (rename, merge, delete), admin only
[x] Static page management (create/edit/delete, PUBLIC or STAFF visibility, draft/published), admin only

User Experience

[x] Self-view profile stats (question/answer/best-answer/comment counts) on GET /users/me/, plus paginated GET /users/me/answers/ and GET /users/me/comments/ for the actual object lists. Self-only, not a public profile, closes out the "Lightweight contributor recognition" item previously listed under Platform Improvements
[x] User search by name/email for admins and school representatives
[x] Responsive design (mobile + desktop compatible)
[x] Loading states
[x] Error handling
[x] Empty states
[x] Form validation

Future Features (Not MVP)

Authentication (Future)

[ ] Email/username and password registration
[ ] Email verification on signup (verification code)
[ ] Email/password login (alongside Google, not replacing it)
[ ] Password reset (forgot password flow via email)
[ ] Password change (while logged in)
[ ] Self-service account management, modification, deletion/deactivation
[ ] Other convenient social login methods (Only if absolutely convenient)

Search (Future)

[ ] Search answers directly, not just questions
[ ] Precomputed search index (stored tsvector column, GIN index)
[ ] Elasticsearch/OpenSearch integration

Notifications (Future)

[ ] Push notifications (for mobile client)
[ ] School submission approved notification trigger (depends on SchoolSubmission existing, see School Data Curation below)
[ ] Notification preferences (per-type opt-out for in-app noise, distinct from the existing email-vs-in-app channel policy, which stays platform-decided, not user-configurable, for MVP)

Moderation (Future)

[ ] Content deletion (moderators)
[ ] User management (moderators)
[ ] Flag capabilities (moderators)
[x] Dedicated escalate-to-admin action, distinct from the general report pipeline (already has a written issue and a backend/frontend split, see the two escalation issue templates)
[ ] Answer/source verification - Allow trusted contributors or moderators to attach supporting sources or verification notes to answers, and indicate when an answer has been verified against an authoritative source.

Internationalization (Future)

[ ] Make the Globe2/country badge on the school profile page permanently visible for every school, including Nigerian ones, once a genuine second-country presence exists. Currently conditional (country !== "Nigeria") specifically to avoid a 100%-redundant badge on a single-country dataset, see the School Directory Fields Display entry in BUILD_LOG_FRONTEND.md for the original reasoning. Marked with a TODO(i18n) comment at the exact line in app/schools/[id]/page.js.

School Reviews (Future)

[ ] Submit school review (Still deciding the details)
[ ] Edit/delete own review
[ ] View reviews per school
[ ] School official response to a review (verified reps only)
[ ] Vote review helpful/not helpful
[ ] Report review (reuses existing report pipeline)

School Verification & Claiming (Future)

[ ] Claim school profile (representative request + admin approval)
[ ] Verified badge on school profile

School Data Curation (Future)

[ ] Curated database of Nigerian universities, polytechnics, and colleges, sourced primarily from NUC, NBTE, and NCCE official lists, cross-referenced against JAMB's institution list and secondary sources. This is the first of what may become several country-specific data pipelines as Academia expands.
[x] Extend the School model with richer fields: institution type (university/polytechnic/college of education), ownership (federal/state/private), state, an official regulatory or JAMB code, source_url, last_verified_at, plus a forward-compatible country field for eventual multi-country expansion. SchoolSourceRecord model added alongside for per-regulator provenance tracking. Schema, serializers, views (three new List Schools filters), and admin all landed. Consuming a real cleaned NUC/NBTE/NCCE dataset via an import command is still pending, tracked separately below.
[ ] Periodic re-verification workflow so curated data doesn't silently go stale
[ ] User-submitted "school not listed" request flow (SchoolSubmission)
[ ] Bulk data import tooling to support the school directory curation effort above
[ ] Admin verification workflow for submitted schools
[ ] School directory/search page filter controls for institution_type, ownership, and state, GET /schools/ already supports all three as query params, no frontend UI built against them yet
[x] Admin school form (SchoolFormModal) gained inputs for institution_type, ownership, state, and country, previously only settable via direct API calls
[x] School directory/search page filter controls for institution_type, ownership, and state, GET /schools/ already supported all three as query params, now wired into app/schools/page.js

SEO & Discoverability (Future)

[x] Site-wide SEO and social share metadata overhaul: root layout defaults, per-page generateMetadata for schools/hubs/questions/tags/ listings/search/homepage, canonical URL strategy (schools now match questions' UUID-primary/cosmetic-slug pattern via lib/urls.js's schoolUrl()), robots/indexing rules (search results noindex when queried), sitemap.js covering schools/tags/questions, dynamic Open Graph image generation (lib/og.js, reusable OgCard) for homepage/ schools/questions/tags, WebSite+SearchAction and BreadcrumbList structured data, and a technical SEO pass. Landed in four phases, see BUILD_LOG_FRONTEND.md.
[x] Platform account visual marker: a single known account id (env- configured, unset by default) gets a small badge next to its name wherever an author is shown, for Academia's own seed/editorial account. Not a role system, no is_admin exposure.

Admin & Moderation UX (Post-MVP)

[x] Collapsible mobile filter sidebar (was previously always-expanded, stacking above results awkwardly on narrow screens)
[x] Search/filter controls added to Pages, Team (moderators/ representatives), Activation Requests, and Unanswered Questions management surfaces
[x] FilterSidebar's mobile badge now shows a real active-filter count instead of a generic dot, wired into HubQuestionList, TagQuestionsClient, and SearchResultsClient
[x] Real server-side search added to GET /questions/unanswered/, UnansweredQueue now searches the full result set, not just the loaded page
[ ] GET /hubs/activation-requests/ intentionally kept as loaded-page- only search in ActivationRequestsQueue, a deliberate choice, not a gap, "bug turned feature"

Monetization (Future. Needs further review)

Every item here must hold to Integrity Over Monetization (project-plan.md): paid tiers affect visibility, promotion, and verification only, never the truthfulness or presence of content students post. - [ ] Verified badge and official review response (already scoped under School Verification & Claiming and School Reviews above) - [ ] Directory placement: a school can pay to appear higher specifically in the school directory browse view, clearly labeled "Featured", never affecting search relevance or question ranking - [ ] Analytics dashboard for school representatives: what students are asking about their school, trending unanswered topics, volume over time, business intelligence that's already a byproduct of the platform, genuinely useful to a school, doesn't touch what students see - [ ] Scoped official announcements: a paying school can pin a clearly labeled banner inside their own hub only (e.g. "Registration closes March 30"), doesn't affect any other school or any organic question - [ ] Careers/classifieds section: paid, clearly labeled listings from employers or recruiters targeting specific schools or departments, a natural extension given SIWES and internships are already common organic topics, doesn't compete with organic Q&A content - [ ] Tiered public API access: free low-volume tier plus a paid higher-volume tier for third-party developers wanting bulk access to the curated Nigerian school directory, monetizes the data-curation work directly (see Public/Developer API below) - [ ] Voluntary support ("tip jar" style), optional, transparent about what it funds, never gates any feature

Public/Developer API (Future)

[ ] API client registration (admin issued keys)
[ ] API key authentication + per-client rate limiting
[ ] Read-only public endpoints (schools, questions, tags)
[ ] Developer documentation portal

Additional Features (Future)`

[ ] AI-assisted search
[ ] Institution-specific document repository (approved forms, timetables, clearance checklists), a natural extension of the Static Pages infrastructure already built, one step beyond a handful of admin-authored pages toward a real per-school document library (new suggestion, worth a real look, since the underlying infrastructure for this already exists and is tested)
[ ] Low-data / offline-friendly mode: a lightweight rendering path for slow or expensive mobile connections, consistent with the mobile-first, data-conscious audience this platform is designed for from the start
[ ] Celery-backed background tasks for email and notification delivery, so a slow SMTP call never blocks an API response

Platform Improvements (Post-MVP)

Backend

[ ] Private "my submitted reports" view for a user's own account, deliberately not part of the public/self profile page, reports are never shown on any profile to avoid making a reporter's activity identifiable to others
[ ] Admin action audit log (who resolved which report, who suspended which user)
[ ] Answer edit history, so a heavily-edited answer's original context isn't lost framed as quality signal, not a leaderboard or engagement mechanic, consistent with Knowledge Over Social Activity
[ ] Refresh-token request deduplication: if multiple requests expire in the same instant, more than one may attempt to use the same refresh token before rotation completes, one could be rejected. Frontend-only concern (Next.js proxy layer), low risk at MVP traffic levels.
[ ] Standalone answer/comment permalink page, so notifications and reports targeting an answer or comment (VOTE, NEW_COMMENT, non-question reports) can deep-link somewhere real instead of rendering as non-clickable text.
[ ] UserSerializer's stats field, currently only attached ad hoc in MeView, should become a proper SerializerMethodField on the serializer itself, so it's included everywhere UserSerializer is used (this also closes a real doc/code mismatch, api-contract.md documents stats as present on the Google Login response, but it currently isn't)
[ ] Google-authenticated users should have set_unusable_password() called explicitly at creation, currently left at Django's blank default, harmless today but worth hardening before email/password auth ships
[ ] GoogleLoginView should explicitly reject login for suspended (is_active=False) accounts with a clear error, rather than issuing a token pair that only fails on the next authenticated request
[x] PATCH /users/me/ should validate full_name isn't blank/whitespace-only and return a proper field error, instead of silently no-op'ing
[x] GET /users/search/ should share the existing "search" throttle scope (60/min) rather than falling back to the general 100/min limit, it's a type-ahead endpoint and will be called more rapidly than that implies
[ ] Google login should verify email_verified from Google's userinfo response before creating/logging in a user, currently trusted implicitly
[ ] Cache invalidation gap: PATCH /schools/{id}/, POST department creation, and PATCH department updates all invalidate school-detail and/or school-list but never school-by-slug, so slug-based lookups can serve stale data (including a deactivated school still appearing active) for up to CACHE_TTL_MEDIUM after a write. apps.hubs already does this correctly for hub-by-slug, apps.schools needs the same fix
[ ] SchoolBySlugView has no admin bypass at all (unlike the by-id route), so a deactivated school is a dead-end 404 via slug even for admins. Worth reconsidering now that reactivation is a real admin workflow and project-overview.md recommends slug as the primary school URL
[ ] Shared "is this request from a platform admin" check duplicated across SchoolListCreateView, SchoolDetailView, and inlined again in DepartmentListCreateView.get(), should be a single helper in apps.core.permissions
[ ] GET /schools/ has no is_active filter for admins, who currently see active and inactive schools in one combined list with no way to narrow to just one, AdminUserListView already has this exact pattern for users, worth matching here
[ ] Several write endpoints have no dedicated throttle scope and fall back to the general 100/min limit: School/Department creation, Answer edit/delete (AnswerDetailView), Mark Best Answer, Comment edit/delete (CommentDetailView), and Static Page creation (StaticPageListCreateView.post). Worth a deliberate pass deciding which of these actually need tighter scoping versus being fine on the general limit, rather than the current case-by-case inconsistency
[ ] Hub deactivation endpoint (admin only), mirroring School's soft-delete pattern. Currently no API path exists to take an active hub offline, is_active on Hub can only ever be set to True through the API today
[ ] If Hub deactivation is added, HubDetailView/HubBySchoolView/ HubBySlugView need the same admin-bypass-for-inactive pattern SchoolDetailView already has, otherwise a deactivated hub becomes unreachable even to the admin who deactivated it
[ ] Cache invalidation gap (combined with the school-by-slug finding from the Schools audit): approving a hub activation request changes School.has_hub but never invalidates school-by-slug:{slug}, so a school's slug-based lookup can show stale has_hub for up to CACHE_TTL_MEDIUM after activation. Worth a shared invalidation helper used by both apps.schools and apps.hubs so this can't drift again
[ ] Unanswered questions queue (GET /questions/unanswered/) only checks ModeratorAssignment, not SchoolRepresentativeAssignment, meaning a rep managing a hub with no moderator assigned yet cannot see their own hub's queue at all. Directly inconsistent with the reasoning already applied to NEW_QUESTION's fan-out (widened to include reps for exactly this zero-moderator-hub scenario), same fix belongs here
[ ] QuestionLockView.check_lock_permission() duplicates hub-role-check logic inline for a third time (Schools and Hubs audits flagged the same pattern already), worth one shared "is moderator or rep for this hub" helper in apps.hubs.permissions instead
[x] Django admin's Question list view doesn't surface is_locked, hard to tell which questions are locked without opening each one
[ ] Search results omit slug (and department, view_count, is_locked) from each question, matching the documented contract exactly today, but meaning the frontend can't build a /questions/{id}/{slug} URL directly from a search result without a follow-up request. Worth deciding whether slug specifically should be added to SearchQuestionSerializer
[x] Search results now include slug (#NN), added to SearchQuestionSerializer so the frontend can build /questions/{id}/{slug} directly from a result; department, view_count, and is_locked remain intentionally out of the search response
[ ] Static Pages endpoints (list and detail) have no caching, unlike Schools/Hubs/Tags which all use the same short-TTL public-read pattern. Deliberately deferred at build time as non-blocking, worth picking up now that it's a formally tracked item
[ ] Tag list (GET /tags/) is fully unpaginated by design, reasonable at current volume but a real scaling risk given tags are free-text and community-generated with no curation. Recommend adding standard pagination now while cheap. This IS a breaking response-shape change ({"results": [...]} becomes {count, next, previous, results}), needs to be scoped as a coordinated backend+frontend issue, not a silent backend-only change
[ ] Tag name (max_length=50) is never validated for length before hitting the database on either write path (TagMergeView's rename branch, and _sync_tags() in apps/questions/serializers.py, the implicit tag-creation path any authenticated user hits via question tagging). A name over 50 characters currently raises an unhandled DB-level error instead of a clean 400. Reachable by any regular user, not just admins, worth prioritizing
[ ] Extend the existing "Admin action audit log" item to also cover tag merge/delete actions, not just report resolution and user suspension
[x] Django admin's Tag list doesn't surface question_count (would need an admin method since it's computed, not a model field)
[ ] Resolving a report with DELETE_CONTENT can orphan other pending reports pointing at content that gets cascade-deleted along with it (e.g. a Question's report deletes it, cascading to an Answer that had its own separate pending report). Already handled defensively (won't crash), but nothing surfaces or auto-resolves these dangling reports
[x] Consider notifying the original reporter when their report is resolved or rejected, matching the "close the loop" pattern already used elsewhere in the notification system, not currently required by any doc but a natural fit
[ ] GET /reports/ only supports a status filter, adding a content_type filter (question/answer/comment) would let admins scope their review more precisely
[ ] generate_unique_slug() has a TOCTOU race: the uniqueness check and the eventual save() aren't atomic, so two genuinely simultaneous creates with colliding names could both pass the check and the second save() would raise an uncaught IntegrityError (raw 500) instead of a clean error. Narrow edge case, but worth a retry-on-collision fix before real concurrent traffic exists
[ ] .env.example documents discrete DATABASE_NAME/USER/PASSWORD/HOST/PORT vars as a fallback when DATABASE_URL is unset, but base.py's actual DATABASES config only ever reads DATABASE_URL via env.db(), those five vars are dead and unused. If DATABASE_URL is left unset while only the discrete vars are configured, the app silently falls back to local SQLite with no warning, a real risk on first deploy. Needs a decision: restore genuine discrete-var fallback logic, or remove the misleading vars from .env.example and document DATABASE_URL as the only path
[ ] STORAGES["default"] in production.py is hardcoded to FileSystemStorage. No media uploads exist yet, but this needs to change to Cloudinary (or similar) before any file-upload feature ships, local disk storage is unreliable/ephemeral on most PaaS hosts
[ ] DRF's browsable HTML API renderer isn't restricted to JSON-only in production, minor hardening, mostly informational exposure rather than a real vulnerability
[ ] DEFAULT_PERMISSION_CLASSES is globally AllowAny, a deliberate and reasonable choice for this platform, but worth keeping in mind as a standing discipline point: any future view that omits permission_classes is open-by-default, not closed-by-default

Frontend

[ ] lucide-react is pinned to ^1.23.0 in package.json, every prior known release of this library uses a 0.x scheme. Not currently breaking anything (icons render, builds pass), but worth confirming via npm ls lucide-react whether this is a genuine 1.0 release or a typo, and spot-checking that no icon names changed if it's the former
[x] TopBar's "Academia" wordmark is plain text, not a link, inconsistent with Sidebar's identical-looking wordmark which does navigate home. TopBar is the persistently visible one, worth making it a Link too
[ ] No centralized z-index scale, ad hoc values (TopBar z-30, sidebar backdrop z-40, sidebar/ProfileMenu dropdown both z-50, NavigationProgressBar z-[100]) risk collisions as more overlays are added. Worth a small shared constants pass before that happens
[x] Minor accessibility gaps in the shell layer: the mobile sidebar backdrop is a bare div with onClick, no keyboard equivalent or button semantics; ThemeToggle's button doesn't expose aria-pressed for its current state. Cheap, low-risk, worth doing together
[ ] Establish a lint or PR-review convention preventing route-shaped files from landing inside components/ (or component files landing inside app/), found two real instances of this during the shell audit (see Project_audit_notes.md), worth preventing recurrence now that contributors are regularly opening PRs
[ ] Both auth-related Route Handlers (app/api/backend/auth/google/route.js's backend fetch, and the token refresh call inside the catch-all [...path]/route.js) have no try/catch around the backend network call itself, only around bad response handling. If the backend is genuinely unreachable rather than just erroring, these throw an unhandled exception that surfaces as a raw Next.js 500 instead of the app's normal error handling. Same gap in two places, worth fixing together
[ ] lib/jwt.js's decodeJwtExp uses Buffer, Node-runtime-only. Not an active bug (these routes default to the Node runtime), but would need atob/TextDecoder instead if any auth route is ever moved to the Edge runtime
[ ] Concurrent-request refresh race, formally consolidated here from BUILD_LOG_FRONTEND.md's Known Deviations: if multiple requests expire at the same moment, more than one may attempt to use the same refresh token before rotation completes, one could be rejected. Unlikely at MVP traffic levels
[ ] RecentQuestionsList's and app/search/page.js's handleLoadMore lack the cancelled-flag unmount guard their own initial-fetch useEffect already uses. Low real-world risk (manual click, not an automatic effect), but worth the same consistency treatment
[x] Question detail's "back to school" link now routes directly via schoolUrl() instead of through the /hubs/{id} redirect hop. QuestionHubSchoolSerializer gained an id field, additive, needed for the link to be buildable at all.
[ ] Meta-row pattern (small icon + label, e.g. author/department/views on question detail, "by {author}" on answers and comments) has been copy-pasted across multiple files with drifting icon sizes relative to their adjacent text (16px icons next to 11-14px text in a few places, now hand-corrected where found during the audit). Worth extracting into one shared component (e.g. a small IconLabel) so icon-to-text proportion stays consistent by construction instead of needing to be independently gotten right in every new usage
[ ] border-gray-200 dark:border-gray-700 (Tailwind's literal gray scale) still appears in place of border-[var(--color-border)] (the project's actual warm-neutral border token) in at least HubQuestionList and app/tags/[name]/page.js's list containers, found independently in two different passes of this audit, likely present elsewhere too. Worth one project-wide search-and-fix rather than continuing to patch individual instances as they're found
[ ] NotificationBell's dropdown only ever fetches once per session (toggle() skips refetching once status leaves "idle"), so both the notification list and its own unread count go stale after the first open, a new notification arriving mid-session won't show up, in the list or the badge, until a full page reload. Worth a deliberate decision on the right fix (refetch on every open, a periodic poll, or something event-driven) rather than picking one blind
[ ] ReportButton's modal uses raw border-gray-300 dark:border-gray-600 instead of border-[var(--color-border)], the only modal shell in the app still on the un-tokenized value, RequestHubModal already uses the token correctly. Small, fix opportunistically
[ ] Filter sidebar (FilterSidebar, shared across HubQuestionList, search, and tag question lists) stacks above the results list on mobile by default grid behavior, meaning a mobile visitor has to scroll past the entire filter panel before seeing any question. Worth a collapsible/toggle treatment on narrow screens, one shared component fix covers all three usages
[ ] Empty states across the app are plain gray text ("No questions yet.", "No comments yet.", etc.) with no visual weight. A small shared EmptyState component (icon + message + optional call-to-action) would read as more intentional and could be dropped in everywhere at once rather than each spot staying ad hoc
[ ] Form inputs, selects, and textareas each independently re-implement the same Tailwind classes (padding, border, radius, background) across roughly 15+ components, with real drift already found during this audit (inconsistent padding, inconsistent border-color tokens). Worth extracting shared Input/Select/Textarea primitives so this class of drift becomes structurally impossible instead of needing to be caught by hand each time
[ ] Error and warning messages are ad hoc
tags repeated across roughly 40 files rather than a shared component. Same drift risk as the input fields above, and would make the already-planned "friendlier user-facing error copy" item trivial to roll out everywhere at once instead of file-by-file
[ ] No persistent "Ask a question" affordance on mobile outside the nav drawer, currently requires opening the sidebar first. Worth considering a small floating action button on narrow screens to reduce that friction
[ ] Muted secondary text (gray-400/500, used extensively for timestamps, meta info, and empty states) hasn't had a real contrast check against the milk-white background. Worth a proper WCAG AA pass with an actual contrast checker rather than adjusting the palette by eye again, given how much back-and-forth the color system already took to get right earlier in this project
[ ] Further markdown editor optimizations to consider, none urgent, worth scoping individually before picking any up:
Undo/redo that understands formatting actions as single steps (e.g. one Ctrl+Z fully removes a toolbar-applied bold wrap) rather than relying purely on the browser textarea's native undo stack, which can split one logical action across several undo steps
Paste handling: auto-converting a pasted raw URL into [text](url) when text is selected at paste time, matching the interaction model of most modern editors
Tab/Shift+Tab to indent and outdent list items (nested lists), currently list items are always flat, no indentation support in either the editor shortcuts or the renderer's allowed output
A visible character/word count for static pages specifically, useful there in a way it isn't for short Q&A content (added to the rich mode attribute)
Auto-closing bracket/marker pairs while typing (typing ** inserts the closing ** automatically with the cursor between them), a genuinely bigger behavioral change worth a deliberate look at edge cases (nested formatting, existing text after the cursor) rather than a quick addition