OmniSvelte v0.2 is here! CLI suite, resource() API, and a cleaner config. Read the announcement
omni — emerald teal
OmniSvelte
Roadmap

Roadmap

The public OmniSvelte roadmap — current status and what's coming next.

OmniSvelte Strategic Roadmap

Consolidated roadmap merging the existing packages/core/README.md roadmap (v0.1–v1.0) with every feature, primitive, and integration discussed in the remote-functions/architecture review of feat/omni-cli. This supersedes the checklist-only version as the working planning document.

Mission

Make OmniSvelte the framework of choice for building anything with web technology — web, mobile, and desktop — by collapsing the repetitive 80% of full-stack work (CRUD, auth, validation, realtime, payments, jobs, forms) into typed primitives, so developers spend their time on the 20% that's actually their product's business logic.

How to read this document

Every feature is scored on four axes so you can sequence work by leverage, not just by what's interesting to build:

Column Meaning
Status ✅ Implemented & verified · 🟡 Scaffolded/stub (types exist, logic doesn't) · 🔧 Implemented but broken · ⬜ Planned, not started · 🆕 New idea from this conversation, not in original roadmap
Value Impact on the mission if shipped — High / Medium / Low
Effort Implementation size — XS (hours) · S (days) · M (1–2 wks) · L (3–6 wks) · XL (multi-month / needs its own design doc)
Primitives Needed The underlying building blocks the feature is built on (see reference table below)
Unlocks What becomes possible because this feature exists

1. Core primitives reference

Most features below are combinations of a small set of building blocks. Get these right once and everything downstream gets cheaper.

Code Primitive Status Notes
RF SvelteKit Remote Functions (query, form, command, prerender, query.live, query.batch) ✅ Integrated in v0.2 via resource(), formSchema(), and fromURL() helpers
MODEL Model ActiveRecord base class + QueryBuilder ✅ Solid foundation; most new ORM features extend this
HOOKS Model lifecycle hooks (creating/created/updating/updated) ✅ Underused — currently no consumer wires anything real into them
SCHEMA defineSchema/fields() + generators (Drizzle, Zod, model) ✅ Single source of truth; should feed forms, docs, and SDKs, not just the DB
PLUGIN OmniPlugin lifecycle interface ✅ Stable and fully typed
VITE Vite plugin codegen/HMR pipeline ✅ Where most "magic" auto-wiring already happens
AUTH Better-Auth integration + generated schema/client ✅ Currently client-reactivity is faked via a Proxy; should eventually use RF
EVENT In-memory event bus (on/off/emit) ✅ No persistence, no queued/retryable listeners
CACHE Cache abstraction (get/set/remember) 🟡 Memory driver only
QUEUE Queue abstraction (jobs, options) 🟡 Memory driver only, no real worker loop
CLI Commander-based CLI (omni) ✅ Strong skeleton; several subcommands are stubs

2. Master feature tables

A. Core ORM & Data Layer

Feature Status Value Effort Primitives Unlocks
ActiveRecord Model + relationships (hasMany/belongsTo/hasOne/belongsToMany) ✅ — — MODEL Everything else in this table
createFactory helper (functional model factories) ⬜ (in old roadmap) High M MODEL, SCHEMA Seeding, realistic demo data, test fixtures
Relationship definitions inside defineSchema (auto-with() loaders) ⬜ (in old roadmap) Medium M SCHEMA, MODEL Removes manual relationship boilerplate per model
Soft-delete query scopes (withTrashed()/onlyTrashed(), excluded by default) 🆕 High S MODEL Safe-by-default deletes; undo/trash UIs
Pagination helpers (offset + cursor) on QueryBuilder 🆕 High M MODEL Infinite scroll, admin tables, API pagination without hand-rolled SQL
DB transactions API (Model.transaction(async trx => ...)) 🆕 High M MODEL Multi-step writes (e.g. payment + order creation) that can't partially fail
Query scopes/macros (reusable named filters, e.g. published()) 🆕 Medium S MODEL Composable, named business queries instead of repeated .where() chains

B. Remote Functions Integration (the API layer)

Feature Status Value Effort Primitives Unlocks
resource(Model) — auto-generates query/form/command CRUD from a Model definition ✅ High L RF, MODEL, SCHEMA Zero-boilerplate CRUD APIs; backend for the v0.3 admin/CRUD UI scaffold
formSchema() — derive a Zod object schema from a Model definition ✅ Medium S RF, SCHEMA One schema definition drives DB columns, server validation, and client form fields
fromURL() — sync URL query params with remote list queries ✅ Medium XS RF Stateful list pages with URL-driven pagination/filtering without boilerplate
getModel() — retrieve a model class by name from the registry ✅ Low XS MODEL Dynamic model access for generic utilities and plugins
omni generate remote — scaffold <model>.remote.ts from a Model ✅ Medium S CLI, RF Lowers the activation energy for adopting remote functions at all
Hook-driven single-flight invalidation (creating/updated/deleted hooks call refresh()/set()) ⬜ High M RF, HOOKS Mutations through the ORM auto-refresh the right cached queries with no manual invalidation code
Remote Function Wrappers ⬜ High M RF Expose custom features, middleware, and config options safely on top of SvelteKit's native remote functions
Auth session as query.live (replacing the client-side Proxy hack in client.svelte.ts) ⬜ Medium M RF, AUTH SSR-safe reactive session state without the current workaround

C. Auth & Authorization

Feature Status Value Effort Primitives Unlocks
Better-Auth integration + generated config ✅ — — AUTH Sessions, social login, etc. (already works)
Policy/permissions layer (can(user, 'update', post), RBAC) 🆕 High L AUTH, MODEL Real authorization, not just authentication — required for almost any multi-user app
Rate limiting middleware (global + per remote function) 🆕 High M RF, AUTH Abuse protection for public forms/commands without per-route hand-wiring
Schema-per-tenant isolation ⬜ (in old roadmap) Medium XL MODEL, AUTH SaaS apps with hard tenant data isolation
Row-level multi-tenancy + tenant resolver middleware ⬜ (in old roadmap) High L MODEL, AUTH SaaS apps with shared-schema tenancy (cheaper to run than schema-per-tenant)

D. Realtime & Live Data

Feature Status Value Effort Primitives Unlocks
WebSocket channels via CrossWS ⬜ (in old roadmap) Medium L new (crossws) Rooms, presence, low-level pub/sub for chat-like features
SSE for lighter realtime cases ⬜ (in old roadmap) Medium M new Cheaper realtime when full duplex isn't needed
Model-level realtime events (Post.subscribe('created', handler)) via query.live 🆕 (implements old roadmap item) High M RF, HOOKS Live feeds, notification badges, dashboards — without standing up CrossWS for simple cases

E. Caching

Feature Status Value Effort Primitives Unlocks
In-memory cache driver ✅ — — CACHE Works today, not production-durable
Redis cache driver + model-level invalidation ⬜ (in old roadmap) High M CACHE Multi-instance deployments, durable caching
`omni cache:clear stats` commands ⬜ (in old roadmap) Low S CLI, CACHE
remember() as a server-side TTL layer on top of remote query dedup 🆕 Medium S RF, CACHE Cheap caching for expensive queries without bespoke per-query logic
HTTP-level cache headers / stale-while-revalidate on prerender/query 🆕 Medium M RF CDN/edge caching distinct from the in-process Redis layer — much cheaper at scale for read-heavy pages

F. Background Jobs & Scheduling

Feature Status Value Effort Primitives Unlocks
Background job queue (BullMQ/in-process) ⬜ (in old roadmap) High L QUEUE Async work off the request path: emails, exports, webhooks
Scheduled tasks (cron-style) ⬜ (in old roadmap) Medium M QUEUE Recurring jobs (digest emails, cleanup, billing runs)
Job status as query.live feed 🆕 Medium S RF, QUEUE Live "your export is processing..." UI with no polling code

G. Storage & Media

Feature Status Value Effort Primitives Unlocks
S3-compatible storage with local-disk fallback ⬜ (in old roadmap) High M new File uploads (avatars, attachments) actually work end to end
form file fields wired to storage driver 🆕 Medium S RF Upload forms (.as('file')) that actually persist somewhere

H. Notifications & Communication

Feature Status Value Effort Primitives Unlocks
Email templates via Resend/Nodemailer ⬜ (in old roadmap) High M new Transactional email actually sends
Database-backed notification persistence + read state 🆕 (closes existing TODO) High M MODEL In-app notification center, read/unread badges
Notifications feed as query.live 🆕 Medium S RF Real-time bell icon / toast feed
First-party SMS drivers for regional gateways 🆕 High M new OTP, transactional SMS for users where email-only auth underperforms

I. Payments, Billing & Generic Webhooks

Feature Status Value Effort Primitives Unlocks
Stripe plugin (subscriptions, one-time, webhooks) ⬜ (in old roadmap) High L PLUGIN, QUEUE International billing
Lemon Squeezy plugin ⬜ (in old roadmap) Medium L PLUGIN Alternative MoR billing
First-party plugins for global gateways 🆕 High L PLUGIN Expands billing beyond Western-only payment processors
Generic webhooks module (incoming signature verification + outgoing dispatch, decoupled from any one payment provider) 🆕 High M QUEUE, AUTH Reusable webhook receiving/sending for any integration, not just payments

J. AI & Intelligence Layer 🆕 (entirely new category)

Feature Status Value Effort Primitives Unlocks
Vector/embedding column type + similarity search (pgvector) 🆕 High L SCHEMA, MODEL Semantic search, recommendations, "find similar" features without a separate vector DB
defineAgent() / structured-output helper tied to remote functions 🆕 High L RF Typed LLM tool-calling and structured generation as a framework primitive, not hand-rolled per project
RAG helper (auto-chunk/embed content on model save) 🆕 Medium L MODEL, HOOKS Drop-in "chat with your data" for any model

K. Local-First & Cross-Platform 🆕 (entirely new category)

Feature Status Value Effort Primitives Unlocks
omni add tauri / omni add capacitor scaffolding 🆕 High M CLI "Build anything with web tech" literally extends to desktop/mobile

L. Business-Logic Primitives 🆕 (entirely new category)

Feature Status Value Effort Primitives Unlocks
defineWorkflow() — state-machine helper for multi-step processes 🆕 High L MODEL, EVENT Order status, approval chains, onboarding flows expressed declaratively instead of scattered ifs
Automatic audit trails (opt-in per Model) 🆕 Medium M MODEL, HOOKS "Who changed what, when" — common enterprise/compliance requirement
Money value type (multi-currency formatting/conversion) 🆕 High S SCHEMA Stops Stripe+Paystack dual-currency billing logic from being rebuilt per project
Regional phone-number/ID validation helpers 🆕 Medium XS SCHEMA Built-in localization for phone and ID formats globally
Feature flags module 🆕 Medium M PLUGIN, CACHE Safe rollout/kill-switches without a third-party flag service
i18n/localization module 🆕 Medium M SCHEMA, RF Multi-language apps out of the box

M. Developer Experience & Testing 🆕 (entirely new category, distinct from the old roadmap's ">90% framework test coverage" goal)

Feature Status Value Effort Primitives Unlocks
Fake drivers for mail/queue/storage/notifications (swap in test mode) 🆕 High M CACHE, QUEUE Apps built on OmniSvelte get testable side effects for free instead of reinventing mocks
Seeded-per-test database helper 🆕 High M MODEL Fast, isolated integration tests without hand-rolled setup/teardown
Remote-function test harness (call query/form/command directly without a running server) 🆕 High M RF Makes the entire API layer unit-testable, which experimental upstream APIs especially need
Dev inspector (Telescope-style panel: recent queries/jobs/remote calls in dev mode) 🆕 Medium L RF, QUEUE, EVENT Visibility into what the framework is doing, replacing scattered console.warn debugging

N. CLI & Code Generation

Feature Status Value Effort Primitives Unlocks
omni init/add/`generate schema migration/db */ui */doctor/install-dependency` ✅ — — CLI
omni tinker — interactive REPL with models pre-loaded ⬜ (in old roadmap) Medium M CLI, MODEL Laravel-Artisan-style exploration without writing a script
Plugin CLI command registration (omni <plugin>:<cmd>) ⬜ (in old roadmap) Medium M CLI, PLUGIN Plugins can ship their own CLI surface
`omni debug:routes models config` ⬜ (in old roadmap) Low S
`omni generate resource auth-page email(currentlyconsole.log('coming soon')`) 🟡 High L

O. UI, Forms & Admin

Feature Status Value Effort Primitives Unlocks
shadcn-svelte integration 🟡 (CLI wiring exists, components don't) High L new Consistent, accessible UI out of the box
Accessible form components with Zod validation binding ⬜ (in old roadmap) High L SCHEMA, RF Forms that are correct and accessible without extra work
Pre-built page layouts (auth, dashboard, docs, marketing) ⬜ (in old roadmap) Medium M new Faster time-to-first-screen
omni generate resource full CRUD UI scaffold ⬜ (announced, not built) High L RF, item B's resource() The single most "wow" demo feature for new adopters
Admin panel (omni serve --with-admin) ⬜ (in old roadmap) High XL RF, MODEL, AUTH Instant back-office for any schema

P. Plugin Ecosystem

Feature Status Value Effort Primitives Unlocks
OmniPlugin lifecycle interface ✅ High — PLUGIN Fully typed and stable
Official first-party plugins: logging, CORS, analytics, error-reporting 🟡 (stubs only) Medium M each PLUGIN Baseline ops needs covered without third-party deps
Plugins can contribute their own remote functions, with conflict detection on query/command names 🆕 Medium M PLUGIN, RF A payments plugin can ship checkout.remote.ts without colliding with the host app
Plugin marketplace/registry ⬜ (in old roadmap) High XL PLUGIN Network effects — the thing that actually makes a framework "the best choice for everyone"
Custom plugin dev framework + scaffolder ⬜ (in old roadmap) Medium L CLI, PLUGIN Lowers the bar for community plugin authors
Plugin dependency management ⬜ (in old roadmap) Low M PLUGIN Avoids plugin version conflicts as ecosystem grows

Q. Observability & Monitoring

Feature Status Value Effort Primitives Unlocks
Structured logging with levels/transports ⬜ (in old roadmap) High M new Production debuggability
Query performance monitoring (omni monitor:queries) ⬜ (in old roadmap) Medium M MODEL Catch N+1s and slow queries before users do
WebSocket connection monitoring ⬜ (in old roadmap) Low S new Operational visibility for realtime features
Observability plugin hooks (tracing, metrics) ⬜ (in old roadmap) Medium L PLUGIN Hook into Datadog/Sentry/OTel without framework-specific glue code

R. Platform Polish 🆕

Feature Status Value Effort Primitives Unlocks
Automatic sitemap/robots.txt/meta-tag generation from routes + schema 🆕 Medium M SCHEMA SEO boilerplate every content-driven app needs, built once
Public typed API/SDK generator (ships an actual client package, not just docs) 🆕 High L RF, SCHEMA A mobile app or partner integration can consume the backend with full types, distinct from "docs from schema"

S. Deployment, Docs & Release Readiness

Feature Status Value Effort Primitives Unlocks
omni build:production ⬜ (in old roadmap) Medium M CLI Optimized production builds
omni deploy --env=<name> ⬜ (in old roadmap) Medium L CLI One-command deploys
Multi-environment config management ⬜ (in old roadmap) Medium M new Safe staging/prod separation
omni docs:generate/docs:serve (API docs from schema + JSDoc) ⬜ (in old roadmap) Medium M SCHEMA Always-current docs with no manual upkeep
Stable semver API, full docs site, migration guides, flagship example apps, >90% framework test coverage ⬜ (in old roadmap) High XL everything The actual "production ready" bar

3. Versioned timeline

Estimates assume continued part-time, largely solo development alongside coursework and competitions — treat dates as relative pacing, not commitments. Each version keeps the original roadmap's scope and folds in this conversation's additions where they belong.

Version Theme Target Headline additions from this conversation
v0.1 — Foundation Done — (no change — already shipped)
v0.2 — CLI & Remote Functions Done Q3 2026 resource(Model), formSchema(), fromURL(), getModel(), omni generate remote, CLI revamp, config migration to vite.config.ts
v0.3 — Auth, Core Services & ORM Next Q4 2026 Auth primitive (typed policies, rate limiting, row-level multi-tenancy); cache, events, jobs/queue, mail, storage, notifications; ORM hardening (transactions, soft-delete scopes, cursor pagination, query macros, model factories)
v0.4 — UI & Forms Planned Q4 2026–Q1 2027 omni generate resource full CRUD UI scaffold, shadcn-svelte integration, accessible form components
v0.5 — Realtime & Live Data Planned Q1 2027 Auth session via query.live, query.live as the lightweight realtime primitive; model-level realtime subscriptions
v0.6 — Payments & Billing Planned Q2 2027 Paystack/Flutterwave plugins, generic webhooks module, Stripe/Lemon Squeezy integration
v0.7 — Deployment & Docs Planned Q2 2027 SEO automation, public API/SDK generator, HTTP cache headers
v0.8 — Plugin Ecosystem Planned Q3 2027 Plugin-contributed remote functions
v0.9 — Hardening & Quality 🆕 New phase Q3–Q4 2027 i18n, audit trails, testing toolkit (fake drivers, seeded DB, remote-fn test harness), feature flags
v1.0 — Production Ready Planned Q4 2027 (unchanged scope — stability, docs, migration guides, flagship apps)
v1.1 — AI & Intelligence Layer 🆕 New phase 2028 Vector/embedding columns, defineAgent(), RAG helper
v1.2 — Local-First & Cross-Platform 🆕 New phase 2028 defineLocalModel(), Tauri/Capacitor scaffolding
v1.3 — Global Business Primitives 🆕 New phase 2028+ Money type, defineWorkflow(), Regional SMS gateways, ID/phone validation

4. Prioritization methodology

Value was scored by how directly a feature removes repetitive, non-business-logic work (the stated mission), not by novelty. Effort was scored by how much net-new design is required versus how much can be assembled from primitives that already exist. Two things consistently pushed a feature up in priority regardless of effort: (1) it fixes something broken rather than something merely missing, and (2) it's a primitive other planned features depend on (e.g. resource() and query.live unlock several other rows in these tables, so they're worth doing early even though they're not the cheapest items).

5. Risks & dependencies

  • Plugin ecosystem has a cold-start problem. A marketplace and community plugins are worthless without a few flagship first-party plugins to prove the pattern — sequence v0.8 accordingly.
  • Bus factor. The roadmap above is intentionally sequenced so that fixing and finishing (v0.2–v0.9) come before expanding (v1.1+), so the framework is usable and credible even if later phases slip.
  • CrossWS maturity — the realtime plan leans on it for full WebSocket support; query.live/SSE are reasonable fallbacks if it proves too heavy.
  • Differentiation vs. crowded competition (Next.js, Remix, Laravel, SvelteKit alone) — the AI-native and local-first primitives are the parts of this roadmap with the least direct competition; they're worth protecting from scope creep into "yet another generic full-stack framework."

6. Success metrics

  • Time-to-first-working-CRUD-app for a new developer (target: under 10 minutes from omni init to a working model + form + list page).
  • Number of production apps built on OmniSvelte.
  • Community plugin count and contributor count once v0.8 ships.
  • npm weekly downloads / GitHub stars as a lagging adoption signal — useful, but secondary to the above.

7. Immediate next actions

  1. ✅ Build resource(Model) for remote-function CRUD generation.
  2. Build Auth & Authorization primitive — typed policies (definePolicy, can, authorize), rate limiting, row-level multi-tenancy.
  3. Build out core services — cache, events, jobs/queue, mail, storage, and notifications with standardized driver pattern.
  4. Core ORM hardening — transactions API, query scopes/macros, soft deletes, pagination helpers, model factories.
  5. Scaffold UI, Forms & Admin (v0.4) — omni generate resource full CRUD UI scaffold, shadcn-svelte integration, accessible form components.

8. Opportunities for Improvement 💡

  • DevTools Suite: A comprehensive local dashboard (similar to Prisma Studio or Nuxt DevTools) running on a separate port during development. It will provide:
    • Database Explorer: Visually inspect, filter, and edit records in your local DB without writing SQL.
    • Remote Functions Panel: Track all incoming query, form, and command calls from the client, including payload sizes and execution times.
    • Background Jobs Monitor: Watch your queues (BullMQ/in-process) in real-time, retry failed jobs, and inspect job arguments.
    • Events & Hooks Tracing: See exactly which model hooks fired during a transaction.
  • First-Class Edge Support: Ensuring the framework and its data fetching primitives adhere strictly to WinterCG standards, enabling deployment to Cloudflare Workers or Vercel Edge seamlessly.
  • Marketplace Tooling: Standardizing the plugin publishing process so developers can safely distribute and monetize high-quality OmniSvelte plugins.
Previous Changelog
Edit this page on GitHub