Roadmap

Honest near-term plans, what is exploring vs committed, and what cartwright will not do.

Cartwright is at v0.33.0 today — the changelog has the full shipped history. The roadmap below is intentionally short and forward-looking. Items are marked [committed], [exploring], or [planned] so readers can tell production-ready from "we are not even sure we want this."

Now shipping

A sample of what's already shipped and live in production — the changelog has the complete list.

Engine v0.11.0 — GEO / AI-discoverability [shipped] — SoftwareApplication JSON-LD + a "Built with Cartwright" block in every shop's llms.txt so AI assistants can recommend the engine, plus agent rules for Gemini + Windsurf. Gated by the default-on, deletable cartwrightBadge flag. See the engine CHANGELOG.
Engine v0.10.0 — 12 feature tracks [shipped] — Resolvable Genome, SEO/GEO Autopilot (Pro), Firecrawl scraper + design-import, Hoptify (parody Shopify design + "import from Shopify" onboarding + Gemini logo generator), GDPR/DSAR, backup, blog, indexing controls, tax/VAT, shipping/fulfillment, and WooCommerce parity. All opt-in, default-off. See the engine CHANGELOG.
/integrations top-nav page — 15 core integrations + 10 Plus-tier MCP integrations with real brand logos and category eyebrows.
/onboarding Plus-tier preview — type a URL, watch the five-agent migration demo, join the waitlist.
/pricing with four tiers — Self-hosted $0, Plus $49/mo, Cloud $199/mo, Enterprise. Resend-backed waitlist for Plus and Cloud.
A2A Headless Merchant foundation — three new endpoints (/api/agent-card, /api/negotiate, /api/escrow/verify), Anchor-Resume negotiation engine, Guardian middleware, P2K scanner. See changelog v3.1.0.
/docs/in-the-box — developer-tone integration walkthrough sitting next to the marketing-tone /integrations.

Now (v0.1 → v0.2)

create-cartwright CLI on npm [committed] — dogfooded on three platforms (Mac, Linux, WSL), then npm publish create-cartwright@0.1.0-beta.
More industry templates [committed] — at minimum eyewear and one home-goods category beyond generic.
demo.cartwright.app live [committed] — a real cartwright shop running in Stripe test mode with nightly DB reset. The single most important conversion asset for visitors.
Auto-rendered /changelog [in progress] — the homepage SoftwareApplication version and the /changelog "current engine release" line now read the engine version straight from CHANGELOG.md at build/ISR time (lib/engine.ts), so the version never goes stale. Full per-entry auto-render from CHANGELOG.md is the remaining step.
next-intl adoption [committed] — messages/en.json + messages/da.json everywhere customer-facing strings exist today.
Video walkthrough [exploring] — 5-minute screencast from npx create-cartwright to first sale. Probably v0.2; depends on bandwidth.

Soon (v0.2 → v1.0)

Cartwright template public [planned, gated] — flip when the contributing milestones are met. Expected 4-8 weeks after Solbrillen.dk launch.
Shopify migration [self-hosted on-ramp shipped in v0.10.0; full agentic flow Plus Q3 2026] — the Hoptify "import from Shopify" onboarding shipped in v0.10.0: with a Firecrawl key it brings your palette (design-import) + products (scraper) across, fail-soft to a demo otherwise. The full five-step agentic migration (customers, deploy) still ships with Plus; live preview at /onboarding.
API auto-reference [planned] — generate the /docs/api/mcp-tools page from lib/tools/registry.ts at build time instead of hand-maintaining it. The current page will drift; this fixes it.
Algolia search [committed, Plus Q3 2026] — confirmed Plus-tier MCP integration. Instant product search wired to the storefront and agent-callable for typed queries. Orama-based search stays for self-hosted forks under ~1000 products.
Plus-tier MCP integrations [committed, Q3 2026] — ten brand integrations shipping with Plus: Klaviyo (email marketing), Mailchimp (email marketing), QuickBooks (accounting), Notion (knowledge), Airtable (operations), HubSpot (CRM), Slack (notifications), Zapier (meta-connector), ShipStation (fulfillment), Algolia (search). All tool-callable from the storefront AI and admin. See /integrations#plus for per-brand pitches.
Multi-tenant docs [exploring] — fork-specific docs hosted on subpaths or via the CLI. Mostly a "would be nice" — unclear if there is real demand.

Maybe later

Headless CMS adapter [exploring] — Sanity / Contentful / Storyblok for the admin Sider CRUD. Likely a separate package, not core cartwright.
B2B / wholesale module [exploring] — tiered pricing, NET-30 invoicing, bulk-order workflows. Big surface, only worth building if a paying user funds it.
EDI [exploring] — only relevant for shops integrating with traditional retail. Unlikely unless a specific fork pulls.

What cartwright will not do

These are explicit non-goals. Listing them here so contributors do not propose them and operators do not expect them.

The template stays MIT, forever. Everything that lands in Teloz1870/cartwright-app is MIT, and so is the public template repo, Teloz1870/cartwright-template. The free, self-hosted path is permanent — you can always run cartwright without paying Teloz a cent. Optional paid tiers (Plus, Cloud, Enterprise) layer hosted services and premium integrations on top; they never gate the core template.
No per-order pricing. Every transaction on a cartwright fork is yours. There are no affiliate-style hooks, revenue-share kickbacks, or volume reporting back to Teloz — even on paid tiers.
No closed-source feature gating of the template. Paid tiers add hosted services and add-on integrations. They do not paywall features inside the template itself. There will not be a "Pro" branch of cartwright with paywalled core features.
No mandatory telemetry. Cartwright forks do not phone home. The MCP introspection surface exists for external AI agents to discover your tool surface, not for cartwright to discover your traffic. Paid tiers add opt-in analytics for your own shop, not for us.

How to influence this

Open an issue on Teloz1870/cartwright-app with the use case — "is this even a real demand?" conversations are welcome there.
Ship it yourself in a fork and document the integration; recipes get pulled upstream when they generalise.